www.piarc.

org
2021R03EN

AUTOMATED VEHICLES
CHALLENGES AND OPPORTUNITIES FOR ROAD
OPERATORS AND ROAD AUTHORITIES
TASK FORCE B.2 AUTOMATED VEHICLES: CHALLENGES AND
OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AUTHORITIES
AUTOMATED VEHICLES 2021R03EN

STATEMENTS
The World Road Association (PIARC) is a non-profit organisation established in 1909 to improve international
co-operation and to foster progress in the field of roads and road transport.
The study that is the subject of this report was defined in the PIARC Strategic Plan 2016– 2019and approved by
the Council of the World Road Association, whose members are representatives of the member national
governments. The members of the Technical Committee responsible for this report were nominated by the member
national governments for their special competences.
Any opinions, findings, conclusions and recommendations expressed in this publication are those of the authors
and do not necessarily reflect the views of their parent organisations or agencies.
This report is available from the internet site of the World Road Association (PIARC): http://www.piarc.org

Copyright by the World Road Association. All rights reserved.

World Road Association (PIARC)

Arche Sud 5° niveau
92055 La Défense CEDEX, FRANCE
International Standard Book Number: 078-2-840606-653-6
Front cover © AdobeStock
AUTOMATED VEHICLES
CHALLENGES AND OPPORTUNITIES FOR ROAD
OPERATORS AND ROAD AUTHORITIES
TASK FORCE B.2 AUTOMATED VEHICLES - CHALLENGES AND
OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AUTHORITIES
AUTOMATED VEHICLES 2021R03EN

AUTHORS/ ACKNOWLEDGEMENTS
This report has been prepared by the working group Task Force “Automated vehicles: challenges and
opportunities for road operators and road authorities” of the Task Force B.2 of the World Road
Association (PIARC).
The contributors (alphabetical order) to the preparation of this report are:

• Martin Böhm (Austria) • Ana Luz Jiménez Ortega (Spain)

• Rudi Botha (South Africa) • Jiwoon Kang (South Korea)

• Domenico Crocco (Italy) • Matt Krech (Canada)

• Luigi Carrarini (Italy) • Eric Ollinger (France)

• Ugo Dibennardo (Italy) • Ian Patey (United Kingdom)

• Jian Gao (China) • Hirotaka Sekiya (Japan)
• Wayne George (United Kingdom) • Martin Thibault (Canada-Québec)

• Matthew Hall (Australia) • Cen Yanqing (China)

• Darina Havlickova (Czech Republic) • Petr Zamecnik (Czech Republic)
• Abdelmename Hedhli (France) • Hamid Zarghampour (Sweden)
• Yuji Ikeda (Japan)
The Task Force B.2 was chaired by Eric Ollinger (France) and Abdelmename Hedhli (France), Ian Patey
(United Kingdom) and Ana Luz Jiménez Ortega (Spain) were respectively the French, English and
Spanish-speaking secretaries.
 EXECUTIVE SUMMARY
2021R03EN

AUTOMATED VEHICLES
CHALLENGES AND OPPORTUNITIES FOR ROAD OPERATORS AND ROAD AUTHORITIES
Vehicle Automation – at SAE levels 3, 4 and 5 – is still in a testing phase in most countries. Many tests
at level 3 have been performed successfully on open roads without any changes to the road
infrastructure and without any observed impacts on congestion. There are various situations which
necessitate that a driver takes control of the vehicle – a fall-back solution that is still allowed at level
3, but not at level 4 or 5.
It is likely that some support from the infrastructure will be needed to reach higher levels of
automation. In particular, well-maintained lane markings and the provision of landmarks appear to be
key features. Automated vehicles cannot, however, rely solely on the physical infrastructure, due to
practical limitations; lane markings deteriorate over time, and it is not practical to forecast when they
become ineffective for automated operation. Therefore, digital infrastructure, including data provided
through high-definition maps and/or through vehicle-to-infrastructure connectivity, is required. Initial
insights on this digital infrastructure have been provided in this report.
Regarding the impact on congestion, two factors need to be considered. The first is the operational
capacity of physical infrastructure. Some studies indicate that capacity is likely to decrease at low
penetration rates and potentially improve only by 2050 or 2055 due to shorter headways. The second
is the change in traffic demand. It has been predicted that within the next 20 years, 60% of the the
world's population will live in cities. If vehicle use is increased without regulation due to increasing
empty automated vehicle trips and changing parking availability, congestion could increase further.
The promotion of automated shuttles could help address this risk.
Some studies also show that the impact on road safety should be positive at high penetration rates,
as already observed due to greater penetration of Advanced Driver Assistance Systems (ADAS). This
assumes that many human factors are considered, such as driver overreliance or loss of driving skills.
The report mentions 11 key challenges to tackle regarding social issues in order to make automation
a success.
Finally, in terms of responsibility, a new balance will need to emerge between the responsibility of the
road operator and the responsibility of car manufacturers (or even the car itself). This balance could
be linked to a certain level of service, but the definition of this level is not yet mature.
AUTOMATED VEHICLES 2021R03EN
2

CONTENT
1 INTRODUCTION ..............................................................................6
2 METHODOLOGY .............................................................................8
3 PHYSICAL INFRASTRUCTURE ......................................................9
3.1 ROAD SIGNS.................................................................................................... 9
3.2 STATIC AND DYNAMIC SIGNS ........................................................................... 10
3.3 LANE MARKINGS: NOT A ROBUST SYSTEM VS GNSS POSITIONING ..................... 13
3.4 NEED FOR LANDMARKS .................................................................................. 14
3.5 ROAD GEOMETRY, INFRASTRUCTURE DESIGN ................................................... 14
3.6 PLATOONING: IMPACT ON THE INFRASTRUCTURE ............................................. 19
3.7 CITY PLANNING ............................................................................................. 20
3.8 MAINTENANCE STRATEGIES ............................................................................ 21
4 DIGITAL INFRASTRUCTURE - CONNECTIVITY .......................... 25
4.1 NECESSITY OF CONNECTIVITY ........................................................................ 25
4.2 USE CASES REQUIRING CONNECTIVITY ............................................................ 26
4.3 REQUIREMENTS FOR CONNECTIVITY ................................................................ 33
4.4 COMMUNICATION MEASURES .......................................................................... 34
5 DIGITAL INFRASTRUCTURE – DIGITAL MAPS AND
POSITIONING ...................................................................................... 36
5.1 INTRODUCTION .............................................................................................. 36
5.2 DIGITAL MAP STANDARDS ............................................................................. 38
5.3 DIGITAL TWIN ................................................................................................ 38
5.4 POSITIONING ................................................................................................. 39
5.5 SATELLITE BLIND SPOTS ................................................................................ 41
6 DIGITAL INFRASTRUCTURE – DATA ISSUES COMMON TO
CONNECTIVITY AND DIGITAL MAPS ................................................ 42
6.1 DATA MANAGEMENT FOR ROAD NETWORK OPERATORS .................................. 44
6.2 DATA ACCESS .............................................................................................. 46
7 IMPACTS ON ROAD NETWORK OPERATIONS (RNO) ............... 52
7.1 IMPACT OF AUTOMATION ON ROAD NETWORK OPERATIONS ............................. 52
AUTOMATED VEHICLES 2021R03EN
3

7.2 IMPACTS ON TRAFFIC OPERATIONS AND ROAD CAPACITY .................................. 52

7.3 PLATOONING ................................................................................................. 56
7.4 AUTOMATED SHUTTLE MUST BE CONSIDERED IN AN URBAN AND RURAL
ENVIRONMENT ......................................................................................................... 58

7.5 HOW RNO CAN HELP CAV ON WORK ZONES/INCIDENTS ................................... 60

7.6 TRAFFIC MANAGEMENT BY RNO .................................................................... 60
8 RESPONSIBILITY AND FINANCING............................................. 62
8.1 RESPONSIBILITY AND INSURANCE ................................................................... 62
8.2 COSTS, BENEFITS AND FINANCING................................................................... 63
9 SOCIAL ISSUES AND THE CONSIDERATION OF HUMAN
FACTORS ............................................................................................ 65
9.1 MAIN PRINCIPAL QUESTIONS FROM THE PERSPECTIVE OF SOCIETY ..................... 65
9.2 THE LIMITATION OF POTENTIAL BENEFITS OF AV BY A RANGE OF HUMAN FACTORS .
.................................................................................................................... 66
9.3 NEW HUMAN FACTORS CAN ARISE ................................................................... 67
9.4 RECOMMENDATIONS FROM THE PERSPECTIVE OF SOCIETY ................................ 71
10 CONCLUSION ............................................................................. 73
11 RECOMMENDATIONS ................................................................ 74
11.1 RECOMMENDATIONS FOR DECISIONS MAKERS............................................... 74
11.2 RECOMMENDATIONS FOR ROAD OPERATORS AND ROAD AUTHORITIES ............ 76
11.3 RECOMMENDATIONS FOR LMICS................................................................. 76
11.4 RECOMMENDATIONS FOR PIARC ................................................................ 77
12 GLOSSARY ................................................................................. 78
13 REFERENCES ............................................................................. 80
AUTOMATED VEHICLES 2021R03EN
4

LIST OF FIGURES
Figure 1: SAE J3016 Levels of Driving Automation [33] .......................................................................... 6
Figure 2 : Examples of variances in Vienna Convention implementations [2] ....................................... 9
Figure 3 : Examples of variances in MUTCD-influenced implementations ........................................... 10
Figure 4 : Variable message sign (VMS) source: MTO .......................................................................... 11
Figure 5 : Changeable message sign (CMS)........................................................................................... 11
Figure 6 : Variable Speed Limit sign (VSLS) ........................................................................................... 11
Figure 7 : Lane Use Management Sign (LUMS) ..................................................................................... 11
Figure 8:Examples of the effect of LED refresh rates - blackened sections.......................................... 12
Figure 9: Magnetic markers and magnetic-induction lines (Source: NILIM, MLIT, Japan) ................... 14
Figure 10: Example of automated low-speed shuttle landmark sign ................................................... 14
Figure 11: Landmark positioning as used in Germany .......................................................................... 14
Figure 12:Pavement marking ................................................................................................................ 15
Figure 13 : Example of continuous hard shoulder (left) and Emergency Refuge Area (right) .............. 17
Figure 14: LIDAR rendering of a snowfall - numerous of purple cloud point (source: Waymo)........... 19
Figure 15 : Bushes visible on street side (Source: NILIM, MLIT, Japan) ................................................ 22
Figure 16: Example of foliage blocking a low-speed shuttle landmark [4] ………………………………………22
Figure 17: Road width in snowy conditions (Source: NILIM, MLIT, Japan) ........................................... 23
Figure 18 : Examples of pavement marking and road sign maintenance issues [14] ........................... 24
Figure 19 : Limitation of range by sensor detection ............................................................................. 25
Figure 20 : Information on-board sensors are not able to detect ........................................................ 25
Figure 21: Data fusion [16].................................................................................................................... 26
Figure 22: Vehicle to Everything (V2X) [17] .......................................................................................... 26
Figure 23 : Road obstacle information provision service...................................................................... 27
Figure 24: Work zone USDOT Project phases (WZDx) .......................................................................... 28
Figure 25 : Congestion information provision service .......................................................................... 28
Figure 26 : Tollgate information provision service ............................................................................... 29
Figure 27 : Merging support service ..................................................................................................... 29
Figure 28: Left Turn Assist (LTA) ........................................................................................................... 30
Figure 29: Vehicle Turning Right in Front of Bus Warning .................................................................... 31
Figure 30: Red Light Violation Warning (RLVW) ................................................................................... 31
Figure 31: Spot Weather Impact Warning (SWIW) ............................................................................... 32
AUTOMATED VEHICLES 2021R03EN
5

Figure 32: Queue Warning (Q-WARN) .................................................................................................. 32

Figure 33 : ITS information services at an intersection [20] ................................................................. 33
Figure 34 : Concept of merging support service ................................................................................... 34
Figure 35: Automated Driving [22] ....................................................................................................... 35
Figure 36: Platooning [22]..................................................................................................................... 35
Figure 37: Remote Driving [22] ............................................................................................................. 35
Figure 38 Level model of a Local Dynamic Map [25] ............................................................................ 37
Figure 39: In-vehicle process to create an environmental model ........................................................ 42
Figure 40: Levels of the Infrastructure Support for Automated Driving [33] ....................................... 43
Figure 41: Examples of ISAD levels along the road network [33] ......................................................... 44
Figure 42: Stakeholders in RNO data exchange [36] ............................................................................ 46
Figure 43: Simulation results representing the percentage change in capacity of each freeway segment
relative to the base scenario................................................................................................................. 54
Figure 44: Impacts on AVs on the freeway capacity in Germany ......................................................... 55
Figure 45:Automated driving of shuttle in rural area (Kamikoani village, Japan) ................................ 59
AUTOMATED VEHICLES 2021R03EN
6

1 INTRODUCTION
In the last few years, vehicles with different levels of automation have been being tested all over the
world. The Society of American Engineers has identified five levels of driving automation (SAE levels):

Figure 1: SAE J3016 Levels of Driving Automation [33]

SAE levels 1 and 2 are driver support features, many have been authorized and deployed in a number
of countries. Levels 3 to 5 represent automated driving features and are still being tested by car
manufacturers with prototypes using different combinations of sensors, Global Navigation Satellite
System (GNSS) positioning and sometimes connectivity.
An important aspect to consider is the Operational Design Domains (ODD) for which a feature has
been designed to operate within. ODDs are defined in the same SAE standard as “operating conditions
under which a given driving automation system or features thereof are specifically designed to
function, including but not limited to, environmental, geographical, and time-of-day-restrictions,
and/or the requisite presence or absence of certain traffic or roadway characteristics.” This includes
roadway types, speed range, environmental conditions (weather, daytime/night-time, etc.), and
prevailing traffic law and regulations.
Road operators and road authorities are keen to understand the impacts of automated vehicles (at
different SAE levels) on traffic management, equipment requirements, road safety, maintenance
strategies, adaptive control and performance optimization of the network and infrastructure design.
This report provides the current state of understanding for all these questions, based on the
experience gathered from testing and experience with vehicle automation all around the world.
AUTOMATED VEHICLES 2021R03EN
7

The report provides an initial explanation of the methodology used for the gathering and synthesis of
the content.
The implications of existing physical infrastructure and its condition on the effectiveness of
automation is discussed as well as the impact of automation on the physical infrastructure.
Irrespective of the quality and condition of physical infrastructure, it is evident that automated
vehicles will increasingly need to rely on a digital twin of this infrastructure, comprising data
transmitted either through connectivity or through digital maps. This is discussed in the chapters
concerned with developing this necessary digital infrastructure.
Also discussed is the impact of automation on road network operations and traffic management, with
implications for road network planning and road agency capabilities.
The closing chapters tackle legal, economic (responsibility, insurance and financing) and social issues
(human factors, including road safety).
AUTOMATED VEHICLES 2021R03EN
8

2 METHODOLOGY
In producing this report, the Task Force firstly defined the main challenges and opportunities that road
network operators and road authorities are likely to face from automated vehicles. An initial workshop
meeting defined the main topics for the Task Force to consider – these topics form the chapters of
this report.
The report begins with a consideration of the implications and responsibilities for road network
operators and authorities on both physical and digital infrastructure. The following aspects of physical
infrastructure have been considered: road signs, lane markings, static and dynamic signs and road
geometry, and city planning and maintenance; including the potential changes to infrastructure design
to facilitate platooning.
Digital infrastructure was identified as a significant area of interest, with consideration given to digital
maps, positioning aspects and data. It was recognised that automated vehicles will impact non-
technical areas such as responsibilities and insurance as well as social issues. The report concludes
with several recommendations including some specific to low- and middle-income countries.
An extensive literature search was undertaken to understand automated vehicle activity – projects,
research, trials, and policy development – across the globe. This was followed by workshop sessions
which identified a list of existing case studies to be utilised for source material - from Australia, Austria,
Canada, China, Czech Republic, France, Italy, Japan, Kenya, Korea, Spain, South Africa, Sweden, and
the United Kingdom. A cross fertilisation session was also held with platooning experts from Europe
to enrich the report during the PIARC World Congress held in Abu Dhabi. A core team of writers was
assigned to each chapter, with reviews by the rest of the Task Force members - to discuss and enrich
the content and perspective. This was followed by a consolidation and editing process to finalize the
report.
AUTOMATED VEHICLES 2021R03EN
9

3 PHYSICAL INFRASTRUCTURE
3.1 ROAD SIGNS
3.1.1 Harmonization of road signs
Road signs are a critical aspect of a safe road network, aligned with the ability of drivers to recognise
and understand them. Human drivers are relatively adaptable in their ability recognise and understand
signs that look familiar even if not exactly as they had expected. Machines are not necessarily as
adaptable. Acknowledging that vehicles are manufactured for global markets, significant international
efforts are needed to harmonize traffic signs and road markings to ensure consistent machine
recognition and safety [1] performance. Due to economies of scale, the automotive manufacturing
industry has limited ability to customize software and sensor system design for each individual
jurisdiction. Therefore, industry-government forums are needed at an international level to support
traffic sign harmonization. The United Nations Economic Commission for Europe has a working Group
on harmonization [2].
There are two major standards for traffic sign harmonization: the US Manual on Uniform Traffic
Control Devices (MUTCD) for Streets and Highways and the Vienna Convention on Road Signs and
Signals. There is also the Southern African Development Community Convention and the Central
American Integration System Convention. Many countries also use individual systems that combine
design principles from both the MUTCD and Vienna Convention. It is also common to have regional
differences in signage within countries and even within sub-jurisdictions as well as on private land.
Many road authorities have national standards for road signs, but regulation and enforcement are left
to local governments. Road authorities that have a national agency responsible for implementing and
approving signage have demonstrated significantly more uniform systems [3].

Figure 2 : Examples of variances in Vienna Convention implementations [2]

Some of the key differences and challenges for harmonization include:
• Units - use of the metric system vs. imperial system; some countries do not indicate units on
road signs
• Shape – use of diamond, pentagonal or triangular signs for various warning purposes
• Colour – use of orange for temporary signage in MUTCD, use of blue for mandatory
information in Vienna Convention vs. supplementary information in MUTCD
• Language – use of text in various languages
AUTOMATED VEHICLES 2021R03EN
10

Figure 3 : Examples of variances in MUTCD-influenced implementations

The benefits from investing in road sign harmonization include improved machine vision system and
algorithm performance as well as limiting barriers to higher levels of ODD implementation and testing.
There are also the benefits of reduced need for costly alternatives (digitization), improved human
recognition and opportunities to improve compliance through more consistent and reliable in-vehicle
alerts.
It is recommended that priority is given to resolving traffic sign issues on expressways/motorways first
as this is where the greatest volume of traffic occurs and where higher levels of automated driving are
most likely to appear first [3]. Harmonization efforts should also focus on sign shape which is the
primary method of recognition as Traffic Sign Recognition (TSR) systems do not distinguish colour well
[3].
Harmonization will take time, and even after a consensus has been reached, physically changing signs
along every road worldwide will be a significant challenge. Meanwhile, vehicle manufacturers and
original equipment manufacturers (OEMs) are introducing different national sign databases in their
systems.
3.1.2 Alternative approaches – adding digitization/connectivity
The Task Force noted that some jurisdictions are taking an alternative approach to prepare for higher
levels of automation. Rather than waiting for harmonization efforts to advance, jurisdictions may
select to digitize all forms of signage by adding infrastructure-to-vehicle (I2V) connectivity or machine-
readable code (e.g., Quick Response (QR) code). The HMI (Human Machine Interface) needs to be
properly defined. With millions of road signs in each jurisdiction, harmonization should be viewed as
a long-term goal. Notwithstanding, there are additional benefits in redundancy from having both
harmonization and digitization of road signage.

3.2 STATIC AND DYNAMIC SIGNS

As noted in the previous section, there are multiple sign standards around the world for static signs.
Most are inspired by either the Vienna convention from 1968 or the US MUTCD first published in 1935.
AUTOMATED VEHICLES 2021R03EN
11

AV sign recognition systems need to read all these signs, whether they are static, dynamic or electronic
signs.
There are four main types of dynamic signs currently in use on roads and highways, built by many
manufacturers to a variety of standards and specifications.

Figure 4 : Variable message sign (VMS) source: Figure 5 : Changeable

MTO message sign (CMS)

Figure 6 : Variable Speed

Figure 7 : Lane Use Management Sign (LUMS)
Limit sign (VSLS)

3.2.1 Traffic Sign Recognition

Traffic Sign Recognition (TSR) is a technology that enables a vehicle to detect and recognize traffic
signs at the roadside e.g. "speed limit" or "children" or "turn ahead". This is an example of an Advanced
Driver Assistance System (ADAS). The technology is being developed by a variety of automotive
suppliers. Vehicle manufacturers are moving towards enabling speed assistance systems and
automated driving systems which use TSR technology; the benefits of successful introduction are likely
to be significant for road safety.
TSR uses image processing techniques to detect and recognize traffic signs. The detection methods
can be generally divided into colour based, shape-based and learning based methods.
An AV equipped with a TSR system can read and recognize traffic signs. Reliability for the reading and
recognition of standard static signs (speed limit, warnings, etc.) by TSR is very high; in Australia and in
France pilot projects have demonstrated a recognition rate of these signs of almost 100%. However,
TSR systems cannot yet read and recognize way finding signs, non-standard and information/advisory
signs.
AUTOMATED VEHICLES 2021R03EN
12

TSR systems rely on signs to be correctly located and maintained (including cleaning) so that visible
light and colour can be captured by cameras.
3.2.2 Traffic Sign Recognition system with dynamic signs
The case studies collated for this report indicate that TSR systems are currently unable to consistently
read dynamic signs (roadside signs or gantry mounted sign). Literature and stakeholder interviews
indicated that the refresh rate of signs and variability of pixel illumination could vary between brands
and designs. Other factors could include the sign size, height and approach angle, as well as level of
illumination from the power source.
3.2.3 CMS, VMS, VSLS and LUMS signs
TSR do not generally recognize Variable message signs (VMS) and Changeable Message Sign (CMS)
systems as they tend to focus on fixed speed sign recognition. Variable Speed Limit Signs (VSLS) and
Lane Use Management Signs (LUMS) also are not recognized by current TSR systems.
3.2.4 Refresh rate
The refresh rate of electronic signs is designed to enable a human eye to see them properly, without
flickering. It is necessary to develop a standard that enables a TSR system to do the same. Different
standards currently exist, with both the New Zealand Standard and EU guidelines recommend that the
frequency of emitted light should be “not less than 90 Hz” while the Australian standard is a
significantly higher 2 kHz.
Flickering in the electronic display may be observed by the TSR systems camera causing it difficulties
in recognizing the sign. This occurs on some electronic signs and not others. In some cases, segments
of the sign may have refresh rates out of sequence with other segments of the sign. Evidence suggests
this flickering effect may not be apparent in direct current powered signs, or signs from certain sign
manufacturers. This could be improved either by improving either the signs or the cameras. As
cameras are getting better year by year the recommended priority would be with camera
improvements.
3.2.5 IVS (In Vehicle Signage)

Figure 8:Examples of the effect of LED refresh rates - blackened sections.

AUTOMATED VEHICLES 2021R03EN
13

A possible solution to the difficulty of detection of fixed or variable signs by TSR systems is to use
infrastructure-to-vehicle (V2I) connectivity. With the help of RSUs (Road Side Units) and OBUs (On
Board Units) in vehicles, the signs can be identified, read and understood by equipped vehicles. This
solution added to TSR systems can improve sign detection, interpretation and response.

3.3 LANE MARKINGS: NOT A ROBUST SYSTEM VS GNSS POSITIONING

3.3.1 Role and challenges of lane markings
Automated vehicle function currently relies on clear, consistent and well-maintained road markings
and signs to navigate. Lane markings are one of the main reference elements for automated vehicles
in maintaining their position on the road and within lanes. However, current lane markings do not fully
meet the needs of automated vehicles. For example, there are various road layouts and situations
(road works, tolling plazas…etc.) where there are no lane markings. The regulations for the secondary
road network in Austria do not require lane markings to be provided. Weather can cause issues, as
some lane markings can be obscured by rain or when the sun is at low angles. To support current levels
of automation, the assessment and maintenance of road markings (including testing and assurance of
retro-reflectivity) is required.
One of the main challenges for road operators is that it is technically impossible to predict or detect
the “failure” of markings (degradation to a point that automation is compromised). It is inevitable that
automated vehicles will have to cope damaged or absent markings, which may need supplementary
information through digital maps. These digital maps will also be required to be maintained, either
manually or through a semi-automated process.
From the information gathered, there is no indication that there is a need for significant change to the
design of road markings or road signs, since the current standards are reasonably well defined in most
jurisdictions. Concerns raised by OEMs predominantly make mention of the degraded state of road
markings and road signs, and not their design.
3.3.2 Lane markings for positioning
It is possible that as vehicles increasingly use other forms of digital infrastructure and mapping to
position and navigate, the issue of road markings may become less critical. However, there is a
challenge that since many existing advanced driver assistance systems (ADAS) rely on road markings,
there may continue to be a reliance on them for some time.
An alternative has been trialled on some automated low-speed shuttles, which needed to consider
reduced availability of GNSS signals in mountainous areas and tunnels, and also with reduced
functionality of sensors under adverse weather conditions such as fog and snow accumulation etc.
Magnetic markers and magnetic-induction lines were installed for vehicles to follow and to identify
their location along sections of defined driving routes. Although this approach requires additional
investment to provide the dedicated infrastructure, both in the road and vehicles, it enabled vehicles
to reliably locate and position themselves along equipped sections. As a result, standards for magnetic
markers and magnetic-induction lines are being examined for some automated driving technologies
to be realized in the early stages in Japan.
AUTOMATED VEHICLES 2021R03EN
14

Magnetic markers Magnetic-induction lines

Sensor sensor
line

Magnetic Marker
Magnetic marker

Figure 9: Magnetic markers and magnetic-induction lines (Source: NILIM, MLIT,

Japan)
3.4 NEED FOR LANDMARKS
Road-side units (RSUs) can also be used to provide landmark information. Static or
Figure 10:
electronic signs with landmark information, such as QR Codes, WIFI or DSRC, could
Example of
be provided to assist automated vehicles. Several countries such as Germany and
automated low-
China are designing new systems utilising landmarks.
speed shuttle
Some AV systems, such as low speed AV shuttles, may have limited localization landmark sign
capabilities and may use sensor data in combination with high-definition maps to
determine the exact positioning along a route (i.e. proximity to a landmark).
Buildings, utility poles, signs or other distinctive landscape features may be used
as landmarks or mounting locations for positioning devices so long as they are
unlikely to be obstructed by changing environmental conditions [4].

Figure 11:
Landmark
positioning as
used in Germany
3.5 ROAD GEOMETRY, INFRASTRUCTURE DESIGN
3.5.1 Sign placement
An important factor for TSR systems is the location, proximity, and applicability of signs from where
they can be viewed – the TSR system needs to read only those signs that apply to the vehicle. Some
field tests have found that TSR systems can read road signs which are not applicable to the road or
lane where the AV may be travelling, such as speed limit signs applying to a separate parallel roadway.
Attempts should be made to ensure that placement minimises confusion, both for automated and
human driven vehicles. Variation in the position of road signs can lead to them being unreadable due
to the distance from roadside and angle of the sign face in relation to the traffic direction. Inadequate
maintenance that results in signs being dirty, misaligned or having reduced retro-reflectivity can also
exacerbate non-ideal positioning.
AUTOMATED VEHICLES 2021R03EN
15

3.5.2 Pavement marking and dedicated lanes

During the early stages of automated driving development, nominated routes should attempt to
avoiding “dynamic” obstacles such as parked vehicles, pedestrians and/or bicycles. To assist
automated driving vehicles in operating smoothly and predictably, measures to suppress on-street
parking, pedestrian conflicts, and bicycles should be considered. Approaches to clearly define the
driving route for automated vehicles (e.g. dedicated pavement marking), and dedicated lanes are
under consideration in some jurisdictions.

Figure 12:Pavement marking

3.5.3 Universal design needs to be upgraded
Universal design requires products to be usable by all people without the need for adaptation or
stigmatizing solutions, and this should now also include CV and AV. Universal design principles need
to be incorporated not only in vehicle design, but also in the design of related infrastructure, such as
electric vehicle charging stations. One of the more common technology adoptions for CAVs is the
provision of necessary information via road-side units. This requires a stable power supply network
and a high-capacity communication network along the all interconnected infrastructure in order to
provide sufficient and steady services to CAVs.
There is a possibility of reducing lane widths in some contexts with the potential for CAV to follow an
established trajectory with lower lateral oscillation than human driven vehicles. More lanes could be
accommodated within the current infrastructure width, increasing the capacity of infrastructure at
some locations. This reduction of the width could be applied in highly constrained locations such as
bridges and tunnels provided that sufficient safety analysis and consideration of all design vehicles
(especially for larger vehicles’ physical widths and swept paths on curves) has been undertaken.
The impacts of some aspects of physical infrastructure design (e.g. road geometry) on vehicle
performance could be improved by AV functions. Automation could smoothly recognize horizontal
alignments and slopes and adjust performance. Sudden changes in road geometry or width and un-
signalized intersections could prove more challenging.
AUTOMATED VEHICLES 2021R03EN
16

3.5.4 Crossings
At-grade crossings are one of the most crucial physical infrastructure locations that could affect AV
and CV design and safe operation. The interaction between pedestrians and AVs will be different to
interactions with human drivers. One challenge is the determination of responsibility between AVs
and pedestrians. Should, for example pedestrians be expected to have an awareness of the presence
of AVs, ensuring they don’t proceed in front of a CAV at the wrong moment; or should the AV systems
be entirely responsible for detecting pedestrians. Practical limitations of onboard detection systems
may result in failure to detect pedestrians under some conditions. Roadside environments can also
have objects and furniture that can obscure the view of pedestrians form a vehicle and detection
systems may not reliably differentiate a human from an object.
• Uncontrolled or random pedestrian crossing locations: Pedestrians are not a “compliant” or
expected part of a system who can be directed and controlled. Pedestrians can be random and
chaotic variables for automation to consider which are vulnerable. Pedestrians as potential
obstructions on motorways are generally quite rare, although exposure can vary in developing
countries, so they still need to be considered for safety. On other road types, the exposure to
unexpected crossing behaviour may be greater and the speed environment may be lower but
implications for automation functions still need to be considered under a wide range of
environmental conditions.
• Zebra crossings: Vehicles are required to stop when pedestrians step onto or are closely
approaching a zebra crossing. It may be expected that pedestrians would wait for an approaching
vehicle to slow significantly before crossing, however this may not always be the case. Although
the priority at a zebra crossing is better defined, automated functions would still need to consider
how to navigate such areas as pedestrians may not perceive the difference between automated
and non-automated vehicles.

• Signalized crossings: A CAV approaching a signalised crossing needs to be aware of the signal
status. It also needs to be recognised that pedestrians may not always wait for the correct signal
phase and may cross late or early fail to be clear of the road when the signals change to a vehicle
phase. Signal phase setting such as the length of red times and flashing amber display where used
could have effects on CAV operation, and some strategies and design changes may be required to
improve clarity.
In areas where there is a reasonable expectation of pedestrians crossing the road, whether there is
formalised control or not, road operators need to consider the potential limitations of AV detection
systems and also the random behaviours that pedestrians might exhibit. This may involve limiting or
clearing some areas of visual obstructions that may obscure the lines of sight near a road where AVs
may operate. Consideration may also need to be given to better segregation between pedestrians and
vehicles, and channelling pedestrians to controlled or clearer crossing locations to improve mutual
visibility and priority for both pedestrians and AVs
3.5.5 Emergency stop areas / hard shoulders
The Transport Systems Catapult in the UK (TSC) has assessed the current situation and the change in
space of refuge areas and hard shoulders. Traditionally, hard shoulders have been provided along
motorways, which provide a continuous strip of hard standing for vehicles to stop in an emergency.
AUTOMATED VEHICLES 2021R03EN
17

However, in more recent years several sections of motorway have been converted to All Lane Running
(ALR) to increase their capacity. The ALR system provides safe harbour areas (referred to as emergency
refuge areas), which are spaces at intervals of up to every 2.5 km.” Some motorways have been
converted to ‘Dynamic Hard Shoulder Running’, which involves retaining the solid white line to
indicate the presence of a hard shoulder and opening the hard shoulder to general traffic or transit
buses during busy times via signs and signals.

Figure 13 : Example of continuous hard shoulder (left) and Emergency Refuge

Area (right)

3.5.6 Different factors why the AV should stop in a secure area

In full motorway / highway pilot mode, vehicles will be travelling at high speeds with the human driver
disengaged from the driving task. It is possible that the driver is not ready to regain control of the
vehicle before it reaches the end of its operational envelope. This could be due to several reasons,
such as:

• Driver falls asleep, suffers some debilitating incident (e.g. heart attack) or becomes otherwise
distracted;
• AV system malfunction or vehicles experience mechanical problems;
• Deterioration of environmental conditions;
• Detection of incidents ahead, such as disabled vehicles in the carriageway, which the AV is
unable to negotiate.
In this situation the vehicle will need a safe area to stop and wait for the driver to be ready, or for
conditions to improve to the extent where the automated control system is able to proceed.
3.5.7 Safe harbours or hard shoulders?
Studies are needed to determine the most appropriate form of safe harbour for AVs, which could also
change over time with changes in automation capabilities and the penetration rates of AVs in the
vehicle fleet. The advantage of a continuous hard shoulder is that there is always somewhere to stop
at short notice. A disadvantage is that hard shoulders in highspeed environments are not a safe place
to stop, especially for extended periods of time. Vehicles travelling in the nearside lane of the
motorway can veer into the hard shoulder due to a lack of concentration by human drivers. A highly
AUTOMATED VEHICLES 2021R03EN
18

undesirable scenario would emerge if a driver has fallen asleep and remains asleep for an extended
period, placing the vehicle and occupant in a highly exposed and dangerous situation.
Safe harbours may need to be appropriately designed and contain enough space for an appropriate
number of vehicles to stop, and regular enough along a route so that AVs can access them when
required. Provisions could also be considered to prevent misuse of these areas recognizing that AVs
could considerably increase the use of such areas.
The frequency and spacing of these harbours is an area that road operators may need to consider. The
distance between harbours will depend on the traffic demand and conditions, and also the type of
AVs that could be present on the segment of road. The ODD will need to consider the availability of
shoulders or safe harbours and the associated risks, which in turn could influence what level of
automation might be achievable over various time horizons.
3.5.8 How sensors work in bad weather conditions
The weather has impacts on road safety which has been clearly demonstrated by historical data.
According to the US Department of Transportation [5], over 5.89m vehicle crashes occur each year on
average, out of which around 1.24m can be attributed to adverse weather conditions including snow,
rain, fog and severe wind. The most common weather-related accidents are due to wet pavement and
rainy conditions which account for 76% and 46% of the accidents respectively.
For the purpose of this report, bad weather is generally defined as wet pavement, rain/hail, snow, icy
pavement, snow/slushy pavement and fog. For AVs as well as for the human driver, the implications
of low friction and the impact on the effectiveness of vehicles sensors are important.
3.5.9 Low friction
In a wet or icy conditions, or on aged pavement, the grip between a vehicle tires and the road can be
substantially reduced. Road operators may need to provide pavement friction information and
potentially dangerous weather conditions through live digital mapping applications and alerts,
although this may be data and time intensive to maintain and keep up to date. AVs needs to be able
to assess the road condition and determine how best to manoeuvre during turns, how best to proceed
from a stopped position and how to best evaluate the required stopping distance. Losing traction
when accelerating or having insufficient distance to safely come to a stop can result in potentially
unsafe driving outcomes.
3.5.10 Sensors
AVs typically use a mix of different types of on-board sensor technologies including, LIDAR, RADAR
and GNSS to continually update their digital awareness and position and to navigate safely in their
surrounding environment. Adverse weather can affect the accuracy and reliability of these sensor
technologies. For example, rain, dirt, leaves or snow can obstruct camera lenses, and the necessary
image processing can be impaired or prevented without clear imaging. This system is less useful if the
lens is not kept clean at all times.
Rainfall and snowfall are interpreted by LIDAR sensors as noise that is superimposed over the
background physical environment. Figure 14 demonstrates the distortion of the surrounding
environment with a sea of purple objects projected around the vehicle due to snow.
AUTOMATED VEHICLES 2021R03EN
19

Figure 14: LIDAR rendering of a snowfall - numerous of purple cloud point (source: Waymo)
A common problem for AVs is designing a system that can operate in bad weather conditions. Falling
rain and snow tends to corrupt sensor measurements, particularly for LIDAR sensors. Very little
research has been published on methods to de-noise point clouds which are collected by LIDAR in
rainy or snowy weather conditions. In a project [7] from the University of Waterloo in Canada,
researchers have identified a method for removing snow noise by processing point clouds using a 3D
outlier detection algorithm. The method is based on the variation in point cloud density with
increasing distance from the sensor, with the goal of removing the noise caused by snow while
retaining detail in environmental features. Some companies (such as Waymo [6]) are also using
machine learning in their sensor system to filter out the snow and see just what's on the road, even if
there are vehicles parked by the curb. It is not clear whether the car can distinguish lanes, but such
system capabilities could be sufficient to avoid collisions.
Recent trials indicate that thorough testing in a range of weather conditions including heavy snowfall,
rain, sleet, fog, smoke, dust, high humidity, heavy winds, frozen road, ground water accumulation or
flow conditions is recommended [4].
Road operators need to understand the limitations of sensor systems and filtering capabilities as some
conditions may remain difficult to resolve for automation and may require additional physical
infrastructure / markers, activities to keep roadways clear of certain obstructions / blockages along
with reliable digital mapping to assist during adverse weather conditions.

3.6 PLATOONING: IMPACT ON THE INFRASTRUCTURE

3.6.1 Effect of increased vehicle capacity on roads and bridges
Some forecasts indicate the automated platoons of vehicles could increase capacity per lane by a
factor of up to 40% when conventional vehicles are excluded from the lane. The shorter the platoon
gap and the greater percentage of penetration, the greater the traffic density and road capacity [8].
AUTOMATED VEHICLES 2021R03EN
20

There is concern on long span bridges carrying closely spaced platoons where the live-load positive
bending moment and shear forces may be substantially greater [9]. There may be a need to revisit
models that assume dilution of heavy vehicles with light vehicles. Aspects of bridge strength that may
need to be reviewed include collision on supports, collision on decks, centrifugal forces on curved
decks and braking forces – increasing the strength of bridges must be done strategically (e.g. connect
key freight areas suitable for platooning) due to the high cost. Areas that are not suitable may be
geofenced accordingly so that platoons dissolve before entering the restricted zone. In some cases,
platooning at distances as close as 4m will not result in damage to existing infrastructure, however it
may be necessary to revisit axle weights and spacing between successive vehicles when developing
policy around the minimum platooning gaps [8].
Pavement fatigue (i.e. rutting) is also a concern, but this may be mitigated by the natural wander of
drivers in Level 1 platoons or programmed random wander at higher levels of automation [10]. Each
jurisdiction will need to revisit their road design parameters when determining suitable road segments
for platooning.
3.6.2 Potential changes to infrastructure design to facilitate platooning
Dedicated lanes could enhance platooning safety since the behaviour of other vehicles would be more
predictable, and the platoon speed would be more consistent. Dedicated lanes for platooning could
also employ more consistent and advanced lane markings which facilitate lateral control and safe
operation at higher levels of automation [8]. Establishing dedicated truck lanes is a potential way to
facilitate platooning with minimal impact on existing infrastructure while also enhancing the safety
and efficiency of the existing traffic flow. These efficiency benefits are likely to be significantly reduced
unless segregated infrastructure is developed [11]. Other highway design changes that could help
facilitate platooning include extended passing lanes, modified ramp acceleration lanes, ramp metering
and wider pavement markings [8].
However, where road capacity is limited, road authorities must also consider dedicated lanes for
public transit needs and public interest objectives must be balanced. The most efficient use of existing
roadways may employ dynamically operated lanes that are customized to meet local area needs; use
of HOV and HOT lanes for platooning could also be considered.
Consideration may also need to be given to the lengths over which platooning may be effective. For
example, highway segments with close interchange or intersection spacing may experience disruptive
lane changing or high frequency of changes to platoons that could impact the efficiency of platooning.

3.7 CITY PLANNING

The prospect of highly automated vehicles presents both challenges and opportunities with respect
to city planning and the way communities grow and develop. The spectrum of private versus shared
AV models may have diverse impacts on various community types: wilderness/remote, exurban,
suburban, inner ring, urban core.
3.7.1 Urban Sprawl versus Densification
In the urban sprawl model, AVs have the potential to make travel more convenient and comfortable
allowing commuters to rest, work or undertake recreational activities and therefore develop an
increasing tolerance to longer commute times.
AUTOMATED VEHICLES 2021R03EN
21

Conversely in the urban densification model, AVs also have the potential to reduce parking needs
either through shared mobility models or automated valet service allowing cities to reclaim valuable
land used for parking lots and garages and repurpose for housing, pedestrian walkways, cycle paths,
green spaces etc.
It should be noted that the urban sprawl and densification are not mutually exclusive and may occur
in tandem with growing population needs [12].
3.7.2 Public Transit
Electric low-speed AV shuttles offer the potential to extend public transit and improve mobility to
areas that are not well served by existing systems. Automated buses on transit ways could potentially
provide cost-effective alternatives to light rail transit. Automated buses operating in a cooperative
platoon would offer rail-like service with narrow lane control and optimized acceleration/deceleration
with significantly less capital cost [13].
It is recommended that road authorities incorporate CV/AV impact analysis into their planning
decisions for new infrastructure and public transit investments.

3.8 MAINTENANCE STRATEGIES

3.8.1 Role of Road Maintenance in Enabling AV Safe Deployment
The condition and legibility of road signs can degrade due to various environmental factors. In addition
to traffic sign harmonization, codes and standards development is needed to baseline inspection and
maintenance practices as well as establish acceptable thresholds for the extent that pavement
markings and traffic signs can crack, fade and deteriorate without compromising machine vision
system recognition. The poor state of lane markings, and signage may hinder CAV deployment in
affected areas [14].
In rural depopulated areas in Japan, roads are typically very narrow, and there can be vegetation along
the roadside. In the example shown in the figure below, bushes were detected as obstacles and the
automated vehicle stopped even though it was not a dangerous situation.
AUTOMATED VEHICLES 2021R03EN
22

Figure 15 : Bushes visible on street side (Source: NILIM, MLIT, Japan)

The effectiveness and accuracy of TSR systems is affected by the visibility of traffic control devices
which can be influenced by rotation, obstruction, deterioration, vandalism, and theft.
In regions with heavy snowfalls during winter, the road width becomes narrower due to piled snow.
Thus, 2-lane 2-way roads sometimes becomes 1-lane 2-way roads. Under such conditions, automated
vehicles must change their trajectories, or must cope with a “give-way to oncoming car” situation.

Figure 16: Example of foliage blocking a low-speed shuttle landmark [4]

AUTOMATED VEHICLES 2021R03EN
23

Figure 16: Road width in snowy conditions (Source: NILIM, MLIT, Japan)
It is also important that roads are maintained to be kept clear of natural objects such as overgrown
foliage or snowbanks as these may be mistaken for obstacles by machine vision systems and cause
AVs to change their trajectories or stop unexpectedly [11]. Landmarks relied on by AV systems (e.g.
low speed autonomous shuttles) must also be placed and maintained to be clear from natural
obstructions [4].
AUTOMATED VEHICLES 2021R03EN
24

Figure 17 : Examples of pavement marking and road sign maintenance issues [14]
3.8.2 AV Use Cases to Improve Road Maintenance
There are also opportunities to use automated vehicles for maintenance use cases such as snowplough
platooning as well as asset monitoring and repair. For example, cameras and machine vision systems
on specialized vehicles can be used to audit signage on road networks [15]. Publicly owned CAVs may
also be able to use their sensors and communication functions to report maintenance issues in real
time to road authorities [14].
AUTOMATED VEHICLES 2021R03EN
25

4 DIGITAL INFRASTRUCTURE - CONNECTIVITY

4.1 NECESSITY OF CONNECTIVITY
“Connectivity” means a connection of a vehicle to an ICT terminal via a network, or to another vehicle.
Current automated driving technology is mainly based on the sensing systems within the automated
vehicle itself, which has many limitations. For example, vehicles alone are:

• unable to effectively detect non-linear line-of-sight and blind spots

• vulnerable to rain, snow, fog, haze and other severe weather
• unable to effectively detect road ice and adjust turning manoeuvres
• limited to a short perception distance
• unable to achieve all-weather automated driving.
The figures below illustrate examples of the ranges within which sensors detect. Sensor recognition
systems function within limits - such as range and resolution of detection - connectivity can
supplement these limits, enhancing and extending the capabilities of the vehicle. Highly automated
driving requires data fusion that integrates various data provided by multiple on-board sensors and
from digital infrastructure.

Sensor Range
Not detected Detected
Fallen object
Wall

Not detected
Sensor Range

Fallen object

Detected

Wall

Figure 18 : Limitation of range by sensor detection

Sensor Range

Detected Not detected

Wall

Figure 19 : Information on-board sensors are not able to detect

AUTOMATED VEHICLES 2021R03EN
26

Data fusion technology, which is defined as the synergistic knowledge from different sources, can
assist in creating the overall understanding of a situation and could be a potential solution to sensor
limitations.

Figure 20: Data fusion [16]

As shown in the figure above, multiple sensors installed in the vehicle enhance detection ability. In
addition, the development and deployment of V2X (vehicle to everything connectivity) technologies
and infrastructure provides vehicles with the ability to communicate with each other, addressing to
some extent the problem of insufficient information that may be available from in-vehicle systems.

Figure 21: Vehicle to Everything (V2X) [17]

4.2 USE CASES REQUIRING CONNECTIVITY

Use cases requiring connectivity from infrastructure (V2I) should be considered separately for high-
speed roads (including expressways, motorways, freeways) and ordinary roads due to the differences
in travel speed, the presence/absence of pedestrians, bicycles, and parked vehicles on roadways.
AUTOMATED VEHICLES 2021R03EN
27

4.2.1 High-Speed Roads

Look Ahead Information (LAI) provision service
The LAI is the information of anticipated events which cannot be detected by on-board sensors. The
LAI includes the information about the road ahead including, for example, accidents, disabled cars,
other obstacles, events, major incident, wrong-way driving, etc.
The National Institute for Land and Infrastructure Management (NILIM), Ministry of Land,
Infrastructure, Transport and Tourism (MLIT) of Japan is conducting public-private joint research for
next generation C-ITS. Regarding LAI, the following information provision services are identified to
realize automated driving benefits on expressways. Agencies in other countries are also testing
connectivity deployments in their jurisdictions which are discussed included the following sections
Road obstacle information
Since current onboard sensors can only detect obstacles for up to c100m ahead only, it is difficult for
automated vehicles to avoid obstacles safely by themselves at highway speeds. With information on
obstacles ahead, automated vehicles will be able to change their routes, change lanes, or apply the
brakes safely.
It also enables the prevention of secondary accidents and the early opening of lanes by quickly
identifying road obstacles. Some car manufacturers operate emergency call service. For example,
when the driver’s air bag is inflated, emergency information is sent automatically to their emergency
call centre. Availability of this information to road operators allows them to respond to an accident
promptly.

■Lane regulation information ■Road obstacles information

(Dissemination of information from damaged vehicles)
Road-to-vehicle communication Road-to-vehicle communication Road
administrator
Lane regulation information Road obstacles information
(Sections, lanes, (Suspected obstacles [early stage], detail of
Road
obstacles, road section [after confirmation
termination schedule, etc.) administrator
by the road administrator])

Road works
Telemarketing Centre

Air bag operation

information, etc.
<Benefit> <Benefits>
*Early response to road *Ensuring the early safety
obstacles of whistleblowers
by automobiles *Early response to road
obstacles by automobiles

Vehicles and automotive management centers

deliver information on airbag operation, etc. to the
road administrator
[Information delivered from broken or damaged
vehicles]

Figure 22 : Road obstacle information provision service

A project on the A2/M2 connected vehicle corridor in the UK tested options for a connected Road
Works Warning (RWW) system which communicates information about upcoming roadworks such as
the location of roadside workers, the configuration of the worksite, and speed limit requirements for
vehicles as they pass the site. It is expected that a system such as this would use a combination of
temporary hardware such as worksite beacons and sensors and permanent infrastructure such as
AUTOMATED VEHICLES 2021R03EN
28

roadside communication devices to communicate data about the site conditions to connected vehicles
[14].
The Work Zone Data Exchange (WZDx) Specification enables road operators to make harmonized work
zone data available for third party use. The intent is to make travel on public roads safer and more
efficient through ubiquitous access to data on work zone activity. Specifically, the project aims to
provide data on work zones into vehicles to help automated driving systems (ADS) and human drivers
navigate more safely.

Figure 23: Work zone USDOT Project phases (WZDx)

The U.S Department of Transportation (U.S. DOT) released a $2.4M Notice of Funding Opportunity
(NOFO) for WZDx Demonstration, which closed on 3rd August 2020. [18] The purpose of this research
program is to increase the safety of the traveling public through the production of consistent public
work zone data feeds across jurisdictions. This funding provides public roadway operators the
opportunity to make unified work zone data feeds available for use by third parties and collaborate
on the WZDx Specification development.
Congestion information
Congestion information provision service sends information to upstream vehicles about traffic jams
(congested section, lane end of congestion, etc.) on interchanges and other exits. This information can
be used by AVs to enable them to change lanes in advance, either to join a queue if they intend to exit
at the interchange or select a free-flowing lane if intending to continue beyond a congested exit.

■Congestion at IC exits
Road
V2I 道路管理者
Administrator
communication Traffic congestion information on IC exits
(Congested areas, end of congestion, etc.)

Road congestion identified

by the road administrator

<Benefit>
*Enabling drivers to respond quickly to traffic congestion

Figure 24 : Congestion information provision service

AUTOMATED VEHICLES 2021R03EN
29

Tollgate information
Current on-board-sensors are not able to recognize with 100% accuracy which tollgates may be open.
Tollgate information provision service ensures reliable gate choice and provides a safer and smoother
passage through a tollgate plaza by providing operational information on each lane.
■Provision of operation information on each lane
Road
Operational information on each lane 道路管理者
V2I communication Administrator

<Benefit> General

*Realization of safe and smooth driving General

within tollgate terminals closure

General

ETC

ETC

Figure 25 : Tollgate information provision service

Merging Support Service
For merging, automated vehicles need to detect traffic conditions on the mainline, find space for
merging, and coordinate their velocity.
Some urban expressways may have relatively short acceleration lanes that may also have constrained
sight lines and limit the ability to manoeuvre in the merge area. In order to support automated vehicles
to merge smoothly, NILIM is developing a system through public-private joint research which provides
information of traffic conditions on the mainline to merging vehicles so that appropriate space to
merge smoothly can be identified as automated vehicles approach by adjusting their speed.

Providing merging vehicles

with information, such as
driving speeds on main-lane
vehicles
Vehicle detection sensor
Road-to-vehicle communication
Sensing speeds and
Generating information to lengths other and so
be provided to merging on of main-lane
ADVs vehicles

Roadside processing unit

Figure 26 : Merging support service

AutopleX in the UK is a project which enhances an automated vehicle's vision and perception, allowing
more efficient and effective traffic merging at difficult lane merge situations and roundabout
junctions. The project is investigating the fusion of external vehicle sensor information with map
AUTOMATED VEHICLES 2021R03EN
30

aligned infrastructure-based sensing (sensing all road users) which can be transmitted to AVs in real
time, along with the applicable road rules context, via standard Internet of Things (IOT) methods [19].
Intersection navigation

• Left Turn Assist (LTA)

An application where alerts are given to the driver as they attempt an unprotected (or filtered) left
turn across traffic (or right turn across traffic), to help them avoid crashes with conflicting traffic from
the opposite direction.

Figure 27: Left Turn Assist (LTA)

• Vehicle Turning Right in Front of Bus Warning

An application that warns transit bus operators of the presence of vehicles attempting to cross in front
of the bus to make a turn as the bus departs from a stop.
AUTOMATED VEHICLES 2021R03EN
31

Figure 28: Vehicle Turning Right in Front of Bus Warning

• Red Light Violation Warning (RLVW)

An application that broadcasts signal phase and timing (SPaT) and other data to an in-vehicle device,
allowing warnings for impending red light violations.

Figure 29: Red Light Violation Warning (RLVW)

• Spot Weather Impact Warning (SWIW)

An application that warns drivers of local hazardous weather conditions by relaying management
center and other weather data to roadside equipment, which then re-broadcasts to nearby vehicles.
AUTOMATED VEHICLES 2021R03EN
32

Figure 30: Spot Weather Impact Warning (SWIW)

• Queue Warning (Q-WARN)

An application that aims to provide drivers timely warnings of existing and impending queues.

Figure 31: Queue Warning (Q-WARN)

4.2.2 Ordinary roads
Providing information on traffic lights through connectivity has been considered at SIP-adus (the
Cross-Ministerial Strategic Innovation Promotion Program: Innovation of Automated Driving for
Universal Services) in Japan. Communication of traffic light sates and impending changes is considered
AUTOMATED VEHICLES 2021R03EN
33

superior to conventional on-board cameras since detecting signal states at 100% accuracy has been
demonstrated to be difficult under various conditions, including backlight.

Figure 32 : ITS information services at an intersection [20]

4.3 REQUIREMENTS FOR CONNECTIVITY

It is necessary to determine which communication methods are to be used from the available methods
as the requirements for connectivity differ for each use case. The following criteria can have different
performance requirements for use cases that utilise connectivity:
• Certainty: Whether information is communicated with full precision
• Latency: Delay in sending and receiving communication
• Data rate (MBps)
For example, in merging support, an automated vehicle on the mainline may be required to accelerate
or decelerate immediately after acquiring information, and thus, a fit-for-purpose communication
method with low latency is necessary. In this example the following steps may be involved to illustrate
the steps in the process and the need for high speed communications.
(i) Sensors detect traffic conditions on the mainline.
(ii) The local server quickly generates merging support information. Information is processed
locally as quickly as possible to enable AVs to adjust their speeds.
Edge processing, which can include the execution of aggregation, data manipulation,
bandwidth reduction and other logic directly on an IoT sensor or device, may be used
because latency of the system is very important. Edge processing is the on-site
processing of data rather than sending it to the cloud or a centralised processing
server. Edge processing can reduce the amount time taken to provide a processed
“answer”, and also reduce the amount of data transmitted by locally calculating the
and providing only the required information.
(iii) The roadside antenna provides information about passing time of mainline vehicles at the
merging point to AVs with low latency and reliable communication methods.
(iv) On receiving the information, AVs adjust their speeds before entering the mainline to
enter the gap in mainline traffic smoothly.
AUTOMATED VEHICLES 2021R03EN
34
6 0 k m/h, 6 m

6 0 k m/h, 5 m

(ii) Generating information locally 6 0 k m/h, 5 m

6 0 k m/h, 5 m

Local server (iii) Delivering information on such matters as the timing of

passing of main-lane vehicles at the merging point
(i) Detecting main-lane (iv) ADVs adjusted their speeds before joining the main-lane traffic
traffic conditions Roadside antenna
Sensor

Merging point

Figure 33 : Concept of merging support service

In addition, road agencies need to take into consideration the costs to deploy connectivity within their
jurisdiction including initial investment cost, communication fees and operating cost, and who will
ultimately bear the cost.

4.4 COMMUNICATION MEASURES

The communication methods used for C-ITS are classified as follows in the TF B.1 report [21].
• Short Range Communications: 5.8 GHz DSRC, 5.9 GHz DSRC or ITS G5 and incoming C-V2X PC5.
The attributes of this category are the short range (geographical distance) it covers, low
latency, capability for two-way communication and small data packet sizes delivered.
• Long Range Communications: Cellular networks including UMTS (3G), LTE (4G) and incoming
5G. The attributes are the long-range scope, low to medium latency, two-way
communications, and larger data packet sizes.
• Wide Area Broadcast: Digital radio (e.g. DAB+) and analogue radio. The attributes are the long
range, medium to high latency, limitation to one-way communication and mid-sized data
packets.
For the merging support system explained above (Figure 33 : Concept of merging support service),
ETC2.0 is adopted in Japan to develop a prototype. ETC2.0 was adopted as the communication method
because the system requires low latency and certainty in communication.
4.4.1 Future communication method
Using 5G technology to achieve collision avoidance and automated driving is similar to using Cellular-
V2X, DSRC, and other methods. At this stage it is expected that 5G will have shorter delays and higher
reliability. 5G will enable high-capacity data transmission facilitating detailed 3D maps and also the
potential to provide remote control of automatic vehicles where authorised.
There are several advanced applications using 5G technology that have been promoted in following
areas [22]:
Automated Driving: the traffic information updates could reduce to less than a minute.
AUTOMATED VEHICLES 2021R03EN
35

Figure 34: Automated Driving [22]

Platooning: Vehicles operate at very close spacing to save fuel and improve the efficiency of cargo
transportation requiring high speed and reliable data transmissions.

Figure 35: Platooning [22]

Remote Driving: when the E2E delay is controlled within 10ms, the braking distance generated by the
remote emergency braking at a speed of 90 kilometres per hour does not exceed 25 cm.

Figure 36: Remote Driving [22]

5G-V2X realized standardization in 3GPP in 2020, which will be followed by a period of a few years of
commercial development.
Industry demand plays an important role in 5G technology research and development processes and
requirements. Internet of Vehicles has become an important application scenario for 5G. 5G
communication technology considers the demand of the automobile and transport industry, and "high
reliability and low delay" has become one of the three major application scenarios defined by ITU and
one of the four major application scenarios determined by China.
AUTOMATED VEHICLES 2021R03EN
36

5 DIGITAL INFRASTRUCTURE – DIGITAL MAPS AND

POSITIONING
5.1 INTRODUCTION
5.1.1 What is a digital map?
A digital map is a digital representation of a physical environment or asset, combining graphical
elements and electronic information to form a virtual representation. They rely on geospatial
information – information that has a geographic or locational component - to locate attributes and
assets. Unlike a physical map that is fixed once created, a digital map can be updated in real time.
Digital maps are built up in layers, using Geographical Information Systems (GIS), where each layer is
dedicated to a type of information, such as roads or contours. Users can choose which layers are of
interest and effectively remove unwanted detail or information to enable them to better read the
maps. As the maps are virtual and electronic, they can be read at any scale. Digital mapping can be
obtained from National organisations such as Ordnance Survey (UK), Institut Géographique National
(France), National Geographic Information Institute (NGII) Korea. These digital maps can be
considered as the “base maps” and are similar to the more traditional paper map, showing a digital
representation of the physical environment. These base maps are useful for wayfinding and navigating
from A to B. Commercially available navigation systems used in vehicles (e.g. satellite navigation or
satnav systems) rely on digital maps to calculate the optimum route to travel; various parameters such
as “fastest route” or “avoid tolls” can be programmed in to these systems.
A more sophisticated level of mapping is the high definition (HD) map which includes additional data
such as lane markings, signs, speed restrictions and road rules [23].
Current generation in-vehicle navigation systems use HD maps and live traffic information, including
social media content, to predict the optimum route. Even these systems rely on a human driver to
correct for any discrepancies within the data set, road network operators do not accept liability for
errors [24].
5.1.2 Why digital maps matter to AV
The safe and efficient operation of an AV relies on:
• Navigation – understanding in real time where the vehicle is in relation to its intended
destination
• Road positioning – guiding the vehicle on the road and in relation to lanes
• Driving in traffic – understanding the relative position of other vehicles and road users and
driving appropriately to avoid collisions.
Whereas sensors provide real-time visibility in the vehicle’s immediate proximity, maps confer a wider
view that allows vehicles to anticipate potential critical situations from the surrounding environment.
The real challenge for HD map generation is the current level of fragmentation in the automotive
industry together with the lack of standards. The concept of a Local Dynamic Map proposed by
Shimada et al [25] illustrates the various types and levels of data and information that connected and
autonomous vehicles may require for safe operation.
AUTOMATED VEHICLES 2021R03EN
37

Figure 37 Level model of a Local Dynamic Map [25]

Navigation – (Types 1 & 2) the operation of an AV relies on the ability of the vehicle’s control systems
to safely navigate from origin to destination on a road whilst interacting with road infrastructure, other
vehicles and road users. Satnav systems rely on digital maps to plot a route and then to track the
location of the vehicle in transit, checking that it remains on the desired route and re-routing when
required – due to a reported incident for example, or where the vehicle takes a wrong turn. In this
navigational process, the vehicle relies on GNSS (Global Navigation Satellite Systems) to match the
vehicle’s position to the digital map. Many navigation system providers update their maps using data
from vehicles using their application via Application Program Interfaces (APIs). One outcome from this
approach is that different providers may use different variations of the base mapping, losing a “single
source of truth”.
Road positioning – (Types 2 & 3) the safe and efficient operation of an AV requires the vehicle (and
its control systems) to understand where it is and to manoeuvre safely using on-board sensors to
detect the environment and “see” features such as road marking and signs. An AV uses a variety of
systems to understand where it is in relation to the road – sensors such as Lidar, digital maps, magnetic
guidelines/markers, and Global Navigation Satellite Systems (GNSS) are used in various combinations.
Driving in traffic – (Types 3 & 4 data) safely negotiating traffic and other road users relies on a
sophisticated array and combination of on-board sensors to understand a highly dynamic and complex
environment rather than a digital map which is relatively static.
Some advanced driver assistance systems (ADAS) and Highly Automated Driving systems rely on
mapping to augment the information from their sensors [26]. It is unlikely that current technology will
enable fully automated driving without using data from digital maps to support or supplement on-
board sensors and positioning systems – this will increase reliance on connectivity to facilitate V2I
communications [23].
AUTOMATED VEHICLES 2021R03EN
38

5.2 DIGITAL MAP STANDARDS

5.2.1 Base Map
The base maps are those provided by the National or State mapping organisations – including
Ordnance Survey (OS) in the UK, Institut Géographique National (IGN) in France, National Geographic
Information Institute (NGII) Korea. Users can subscribe for automatic updates to the base map. Base
maps themselves are not useable by AVs for navigation – unlike a human map reader – and rely on
their conversion to the relevant digital standard that the AV can “read” and use. Base maps are used
by commercial organisations (such as navigation and wayfinding providers) to produce their own
mapping systems for use in vehicles.
Different standards for digital mapping are used in different countries – the European Commission’s
INSPIRE programme aims to create standardization for mapping [27].
5.2.2 Ownership
Taking the UK example again, the ownership of the base map belongs to OS.
5.2.3 Changes to the base map
Permanent changes are provided by the originator of the map, as mentioned above this can be
accessed via subscription. Dynamic or temporary changes can have a significant impact on the
operation of AVs, such as road works, changes in lane use (e.g. a bus lane) etc. since they are not
included in the changes to the base map and require either a Digital Twin or a High Definition map. An
important consideration for road operators and AV users/operators is the availability of proposed
changes to road layouts and how quickly they can be implemented to the map so that the AV
recognizes the change as soon as it happens [28]. Most current navigation systems do not include
automatic updates to maps which can be frustrating for human drivers but not generally unsafe,
however real time synchronisation between digital and physical information is essential for the safe
operation of AVs. The European Commission TN-ITS INSPIRE [29] programme aims to provide a
platform to enable road network operators to publish changes to their networks so that map makers
and service providers can update their maps and make them available to AVs.

5.3 DIGITAL TWIN

5.3.1 Definition
A digital twin is “a realistic digital representation of assets, processes or systems in the built or natural
environment “ [30]. In the context of a road network, a digital twin is built upon a digital model of the
road and all its attributes – signs, markings, structures, drainage, fences, etc. Rather than being a static
model, a digital twin is linked to the physical asset using live data flows from sensors which input
current performance data from the physical twin and feedback into the physical twin via real-time
control. Therefore, the digital and physical are twinned in real time.
Digital twins can exist at any scale, from a street to a city to a national road network.
Digital twins are useful for road network operators when designing, constructing, and maintaining
their networks, however it is unrealistic to expect digital twins to be created as a retrospective
exercise; the cost of digitising existing networks would be prohibitive. As the use of digital tools
becomes commonplace new and upgraded roads will increasingly have a digital twin.
AUTOMATED VEHICLES 2021R03EN
39

5.3.2 HD Maps
HD (High Definition) maps are like digital twins in that they contain much more information than a
standard digital map – lanes, signs, road markings, speed limits and other traffic restrictions. HD maps
make available layers of information and detail to support and enable AV operations. Commercial
organisations have created products to develop and sell HD mapping. The Austroads project team was
unable to find commercially available HD maps of Australia or New Zealand in 2019 [23]. The
researchers concluded that the lack of commercially available maps was likely due to a lack of market
demand.
The use of HD maps and digital twins is an emerging area and the standards and responsibilities for
generation, owning, updating, and publishing are also emerging.
5.3.3 Interaction with AV
A digital twin will improve the reliability of AV operation by enabling a digital interaction between the
information contained within the twin and the AV, rather than relying on sensors to read, process and
understand information on physical signs and markings. As the penetration of AVs increases towards
100%, the role of signs and markings for traffic control becomes increasingly redundant. This does not,
however, apply to pedestrians, cyclists, scooters, and other modes and would require further
investigation into the management and operation of a multi-modal transport network with varying
levels of automation across modes.

5.4 POSITIONING
5.4.1 Satellite positioning
Global Navigation Satellite Systems (GNSS) are used to determine the location of a vehicle fitted with
an appropriate receiver, using signals from satellites. Satellite Navigation has a central role in the
technology-mix that is currently needed to perform the level of positioning required by automated
vehicles; in a recent inquiry made in EU among pilot projects, it resulted in all of them are making use
of GNSS as well as fusing GNSS positioning with several other on-board sensors. Since GNSS, as well
as other on-board sensors, is affected by external conditions, it is essential that a high quality GNSS is
used; this requires optimisation of some key parameters such as availability, integrity, accuracy,
sensitivity and robustness to interferences by exploiting all the new GNSS signals/services and recent
receivers’ development (e.g. dual frequency).
GNSS is an established technology providing absolute positioning, irrespective of weather conditions.
GNSS is independent of any sensors based on perception and may resolve any ambiguities from
sensors. It provides accurate timing information, which is needed for sensor fusion, to synchronize the
output data of the on-board sensors. Although GNSS could function stand-alone for the less
automated functions, there is a continuous influence of environmental conditions and local errors.
Therefore, GNSS working in a complementary and interoperable manner with other automotive
technologies is essential in all the levels of automation defined by the industry.
Space Based Augmentation Systems (SBAS) are also used to deliver the positioning requirements for
AV [28].
AUTOMATED VEHICLES 2021R03EN
40

Galileo
In Europe, policy makers understood that Galileo, the European GNSS, could better support
automated vehicles for cooperation between vehicles and for better positioning. The 2016 Declaration
of Amsterdam worked to address these barriers and developed a shared European strategy on
Connected and Automated Driving. Most recently, the European Parliament adopted the European
Strategy on Cooperative Intelligent Transport Systems (C-ITS) in March 2018 in which Galileo also has
a critical role.
Australia & New Zealand
Free access to an SBAS is not available in Australia nor New Zealand [28] which has the potential to be
a barrier to the operation on AVs. The solution could be a hybrid system using GNSS and other, ground
based, positioning technologies. The result would be a requirement for different hardware on vehicles
in these markets which may not be commercially viable.
5.4.2 Level of accuracy required
The level of accuracy required for automated driving depends on the level of automation:
LEVEL 2 - Partial automation:

• The positioning requirement of basic ADAS functions is at metre-level accuracy (in the range
of 2 – 10m)
• The GNSS solution meets these requirements complemented by inertial navigation, odometry,
and dead reckoning.
• Type of GNSS solution: normally single frequency receiver (recently multi-GNSS)
• LEVEL 3 - Conditional automation:
• The positioning requirement of Advanced Cruise Control functions is at half metre-level
accuracy (in the range of 40 – 50cm)
• The GNSS solution meets these requirements complemented by radar and ultrasound sensors,
eHorizon, Computer Vision and Simultaneous Localization and Mapping (SLAM)
• Type of GNSS solution: from single to double frequency, more accurate code measurement
augmented by Differential GNSS (DGNSS) and SBAS.
LEVEL 4 – High automation:

• The positioning requirement of Autonomous Driving functions is at decimetre-level accuracy

(in the range of 25 – 45cm)
• The GNSS solution meets these requirements complemented by similar sensors of Level 3 and
by 3D Maps, V2V and V2I communication.
• Type of GNSS solution: from double frequency to triple frequency, high accuracy carrier phase
measurements including Real-Time Kinematic (RTK) or Precise Point Positioning (PPP), or a
combination of both, and authentication encryption
LEVEL 5 – Full automation:
• The positioning requirement of fully Driverless functions is at decimetre/cm-level accuracy
• The GNSS solution meets these requirements complemented by similar sensors of Level 4, and
by machine learning and artificial intelligence.
AUTOMATED VEHICLES 2021R03EN
41

• Type of GNSS solution: as level 4, but guaranteeing more robustness from multiple frequency,
enhanced authentication, better availability and very efficient high accuracy / high integrity
solution.
5.4.3 SF-PPP
Single Frequency Precise Point Positioning (SF-PPP) uses a low-cost receiver with a single frequency,
single antenna and single GNSS constellation to provide greater levels of positional accuracy for AV.
This has been shown to achieve accuracy of 50cm [31]

5.5 SATELLITE BLIND SPOTS

GNSS positioning relies on line-of-sight from the vehicle to the satellite(s), therefore when a vehicle is
within a tunnel the link to the satellite is lost and the vehicle is no longer able to communicate and
establish its position. There are various technical solutions designed to overcome the problems within
tunnels to maintain positioning information whilst the vehicle is out of sight of satellites. Thus, GNSS
is normally assembled with an Inertial Navigation System (INS), map-matching and dead reckoning
techniques. The navigation in tunnels is also supported by the information provided by radar, video
sensors or differential corrections from terrestrial stations technologies.
The same problem occurs in some city situations where high-rise buildings create “canyons” where
satellite signals cannot reach vehicle-based receivers, driving within multi-story car parks or basement
car parks presents the same problem. In this case, the use of multifrequency GNSS signals can be an
appropriate solution to substantially reduce the consequences of reflection in buildings (the so called,
“multipath effect”) that distorts the positioning calculation.
AUTOMATED VEHICLES 2021R03EN
42

6 DIGITAL INFRASTRUCTURE – DATA ISSUES1 COMMON TO

CONNECTIVITY AND DIGITAL MAPS
Digital Infrastructure is the digital representation of the road environment required by Automated
Driving Systems. It includes the connectivity between the vehicle and the road infrastructure, the
positioning infrastructure, digital maps and links to advanced Road Management Systems. [32]
The role of digital infrastructure is to provide:

• Support for internal procedures

• Services to external users, including traveler information services
• Support to connected and automated vehicles.
It is expected that automated vehicles as well as automated functions supporting in-vehicle ADAS will
need to be supported by additional digital data and services provided by Road Network Operators. In
this respect it is important to understand the basic logic of ADAS functionalities. A vehicle uses data
from several in-vehicle sensors (including cameras, lidar and radar) and merges them with external
data, e.g. map data provided by a map-service-provider or data received via C-ITS interfaces. In a so-
called “perception phase” all aggregated sensor data are used to build objects resulting in a situational
analysis. All data is interpreted to create an “environment-model” which forms a basic requirement
for the in-vehicle decisions to activate or deactivate specific automated functions. If automated
functions are activated, the “environmental-model” is aligned with the Operational Design Domain
(ODD) where automated functionalities are activated.

Figure 38: In-vehicle process to create an environmental model

In the context of the “perception phase”, data from road network operators are highly desired to
support automated driving tasks. In addition to the basic data requirements in Europe, a discussion is
currently ongoing regarding support for automated in-vehicle functions in case of events, such as poor
weather conditions, roadworks, or other incidents on the road network. The so-called “Infrastructure
Support levels for Automated Driving” (ISAD) focus on mixed traffic conditions and define the level of
support for ADAS as well as for non-equipped vehicles to enable all vehicles within the road network
to be supported with respective services. These levels can be used to classify the road infrastructure

1
See also PIARC report “Road related data and how to use it” published in December 2020
AUTOMATED VEHICLES 2021R03EN
43

for the supported vehicles, both automated as well as conventional. A “level E road” provides no
infrastructure support, whilst a “level A road” provides full support for ADAS functions via digital
information and by physical infrastructure such as variable message signs (VMS) for conventional
vehicles. It is expected that, based on the information received from the road network operator, a
vehicle will decide to activate/deactivate an automated driving function.

Figure 39: Levels of the Infrastructure Support for Automated Driving [33]
It is evident that not all roads will provide the same range or quality of services to road users, different
segments along a road will provide different support levels. Bottlenecks and safety critical road
stretches and junctions will likely provide greater AV support, while more traditional support will likely
be given along rural and remote roads and in peripheral areas. Such scenarios also reflect the current
situation, where a better physical infrastructure can be expected, for example on motorways than on
rural mountainous roads in woodlands. Figure 40 shows, what such an ISAD road classification may
look like. The ISAD level information for all road stretches needs to be described to provide the
operating environment information to in-vehicle systems and associated service providers.
Accordingly, drivers would be informed what services they can expect on what road-links.
AUTOMATED VEHICLES 2021R03EN
44

Figure 40: Examples of ISAD levels along the road network [33]
The following sections further expand on the detail of the digital elements needed to support ADAS
functions by road network operators, specifically in relation to the processes for data processing and
data management.

6.1 DATA MANAGEMENT FOR ROAD NETWORK OPERATORS

If a road network operator intends to provide digital information into vehicles, proper processes need
to be established in addition to the technical systems and equipment required. PIARC Task Force B.1
published a technical report on “Big Data for Road Network Operations” [21], describing the
framework for big data management and usage based on global best practice. Specifically, big data
management deals with capabilities to process big amounts of data and to perform proper data
analytics using Artificial Intelligence (AI) and machine learning techniques. To facilitate the setup of a
digital infrastructure, skills on big data, AI, Internet of Things (IoT) etc. are highly beneficial to road
network operators to provide accurate and trusted information into a vehicle. Setting up proper data
analytics procedures will result in highly valuable services supporting ADAS functions. The provision
of services with high quality timely content, with accurate positioning information (even with lane
specific detail) will undoubtably be used in vehicle systems.
6.1.1 Requirements on data
There are mainly two kinds of data to be used by automated vehicles, static and real-time. Static data
provides a general description of the road characteristics as well as the legal requirements, (e.g. road
rules and general user priorities for making decisions). Real-time data provides accurate information
on current road conditions, incidents as well as current active regulations (e.g. variable speed limits
or dynamic lane restrictions). "Known quality” is a fundamental requirement for both types of data.
Smart data is defined as structured accurate data with known benefits that serves specific use-cases.
The ownership and basic resources of data used to support ADAS functions need to be resolved. It is
important to know where data comes from and who owns data provided. Only if the resource is known
and validated can data be used for digital infrastructure support.
AUTOMATED VEHICLES 2021R03EN
45

For the data itself a structured dataset is of high importance, where raw data are structured in a way
that spurious outliers are eliminated. False data needs to be eliminated, and only trustworthy data
used in the dataset that forms the basis for RNO services supporting ADAS functions.
The purpose of original data collection also needs to be understood, especially when it is possible that
collected data could be reused for other purposes, e.g. to support ADAS functions. Data collected from
a road junction (i.e. intersection) for traffic planning issues may have an inherent aggregation level
within the collection process which cannot be easily disaggregated. The quality of such data may not
suit the needs for other uses that require more disaggregated information (e.g. where lane specific
information is needed). As a result, data that has high quality for one purpose might have low or even
no quality for another use case. For the support of ADAS functions, it is important to be clear on the
potential utility of pre-existing data, especially regarding safety, efficiency, environmental guidance,
or law enforcement.
For the reasons outlined, a corresponding and clear explanation of datasets is needed, referred to as
metadata-information. Metadata ideally describes the content and structure of data as well as the
primary purpose for which data has been collected. In this regard the European Commission has
published the DCAT-AP (Data Catalogue Vocabulary – Application Profile) specification for Metadata
[34], which provides guidance to describe datasets in a harmonised structure. That specification is
used as basis for data access at European Access Points for transport related data [35].
6.1.2 Data processing and data management
Data is the basis to enable new services, such as those for automated vehicle support services.
However, raw data has only a marginal benefit or limited useful value by itself. Usually data needs to
be processed to create new information or combined with other complementary data to improve
existing services. This service generation by a road network operator level is done in integrated,
holistic traffic management centres. Therefore, it is highly important to set up internal processes as
well as technical expertise to process data in real time to enable highly reliable and quality support to
ADAS functions within equipped vehicles.
In setting up proper data processing mechanisms it is important that the resulting services will not
only serve automated vehicles, but various interfaces to internal and external stakeholders that need
to benefit from improved service quality. These services with improved quality will not only support
internal asset management and planning issues but also assist external stakeholders that can benefit
via a broad range of services, as shown in Figure 41. This provision of services via different interfaces
needs to be supported by proper data management structures where both internal and external users
need to have access to the same data stack through appropriate agreements and controls. Such
infrastructure needs to avoid the duplication of data or the provision of similar data with opposing
content and with different quality.
AUTOMATED VEHICLES 2021R03EN
46

Figure 41: Stakeholders in RNO data exchange [36]

Within Figure 41 the focus is on data and service provision from road network authorities towards
external stakeholders, including connected and automated vehicles. The traffic management centre
(TMC) (or traffic control centre (TCC) as shown in the figure), operated by a road network operator,
processes all data and provides access to the data as well as to improved services via different
interfaces. All external stakeholders are served by the same, central system. In this example, data can
be provided via National Access Points (NAP) (Option 1), which is a European obligation to be set up
in EU-countries. ITS Service Providers can access this data via the NAP., Alternatively, the NAP may
only provide a catalogue type service based on metadata which re-routes the request to the road
network authority, in which case data access could be given directly by the road network operator
(Option 2). An ITS Service Provider could then use the data received for their own services, e.g.
traveller information or navigation services.
In addition, the TMC (or TCC) can also exchange data and services with connected vehicles using their
own C-ITS infrastructure. C-ITS services can rely on direct short-range communication or via cloud
services for long-range communication using cellular networks as an example. It is important to
reiterate that whatever interface is used, data exchanged or transmitted always needs to come from
a central data management structure set up at the TMC (or TCC) of the road network operator.
It is also expected that data collected by vehicles will also be accessible for use by road network
operators. Single vehicles could serve as moving sensors improving data and service quality of the
road network operator. This data exchange and access to data opens several additional questions
regarding data access in general and specifically on data usage, legal aspects, and cost of data.

6.2 DATA ACCESS

To enable the proper usage of data for traffic management purposes, as well as for the creation of
services to support ADAS functions, access to data is crucial. The use of standardised data exchange
interfaces, based on universal standards, is highly recommended. Such standardised interfaces enable
easy access and use of data without intermediate complex conversion processes, as during conversion
processes data elements and critical knowledge might be lost. Proprietary interfaces are mainly used
to strengthen a direct dependency between a data-provider or component supplier and a data-user.
AUTOMATED VEHICLES 2021R03EN
47

However, in a quick moving world the use of standardised interfaces is highly recommended to ensure
flexibility in terms of business-relations, even after the installation of components or the
establishment of cooperation. Additionally, further data providers that contribute to new levels of
quality, or the creation of even new services, can easily be added by using a standardised data
exchange interface.
Alongside the technical aspects of data quality, access and use, several other issues related to business
and legal aspects need to be considered, when discussing data access and use. Liability issues, in
particular, are developed in [Austroads Report (AP-R581-18) Section 2.1]
6.2.1 Willingness to give access to data
Data has a monetary value. In principle all actors need to be sensible when sharing data and giving
access to third parties. Usually, data sharing is based on cooperation models between different actors
that clarify the content of data, based on metadata descriptions, and the commercial contract details.
Prior to further discussion of cooperation models, data access principles relating to different actors in
the domain of automated driving need to be discussed and agreed.
Data owned by road network operators
In many cases road network operators are public entities and need to follow national regulations on
data access. In many areas, public data are treated as existing for the public good generally resulting
in an open data policy. In Europe, for example, several European regulations define the access to data
owned by public entities. Sometimes there are only very generic regulations for all kinds of public
data, but there are also regulations in the transport sector defining data categories that need to
become publicly available. These regulations are aimed at giving access to public data by private
companies for the generation of new services. So, usually road network operators are obliged to give
access to their data based on a marginal cost principle. This marginal cost principle can also be used
for service provision by the road network operator.
Currently most road network operators are not willing to provide any guarantee (thereby limiting their
liability) for any supplied data [37]. This is especially the case as current legislation for safe road
network operations (in most if not all jurisdictions) still considers the physical infrastructure, including
signs and road markings, to be the regulatory element. The need for operating a digital infrastructure
that provides regulations in a digital machine-readable format, including the digitalisation of the
national road traffic acts, may be required for higher automated in-vehicle functions. In this regard
standardisation bodies have already started to work on the management of electronic traffic
regulations (METR); liability issues towards road network authorities might arise in the future.
Data owned by private actors
Private actors associated with automated vehicle operations include vehicle manufacturers, telecom
operators, digital map providers/operators, and other C-ITS service providers. Usually, private actors
are not directly affected by national legislation on data provision. This is due to data and services
provided by private actors being incremental parts of their business development and product
offerings. Therefore, private actors are in principle very careful in allowing access to their data. This
can prevent competitors from providing the same service quality, allowing a product developer to
maintain a legitimate market advantage based on their investment. Few companies are likely to be
AUTOMATED VEHICLES 2021R03EN
48

willing to share their data – to avoid their competitors gaining insights into their current development
status.
For automated vehicles, data exchange of specific types between different actors and vehicles of
different brands will be a pre-condition for their successful deployment. Vehicle-to-Vehicle
communication, where one vehicle informs the other on possible dangerous situations, is highly
relevant and requires a level of collaboration and data exchange. Currently it is not defined what kind
of data will need to be exchanged between different vehicles for a normal driving task, and what data
will be generated by in-vehicle systems. For vehicles travelling at high speed in a mixed vehicle
environment, vehicle-to-vehicle data exchange will be of high importance to enlarge the operating
horizon of a single vehicle. Current in-vehicle sensors provide an overview of a vehicle’s surroundings
up to approximately 300m ahead. For a vehicle travelling at 120 km/h, it would potentially have a
reasonable overview of the upcoming conditions within the next 9 seconds, depending on traffic and
environmental conditions. The relative shortness of this period might result in sudden uncontrollable
movements that could be experienced in non-harmonised traffic flows. Exchanging data between
single vehicles would drastically enlarge the potential operating horizon enabling safer vehicle
operation and more efficient traffic flows.
Road network operators would benefit from data generated by private actors. If, for example, static
geometry data provided by road network operators are not accurate enough, a feedback loop from
private actors concerning an alteration to the geometry would help all actors in the domain. Dynamic
data related to efficiency or safety are of high relevance for road network operators. Privately
generated data could help operators to improve maintenance processes. Currently data of this kind
held by private actors is hardly accessible for reasonable costs, mainly based on business issues, where
other private actors might also unfairly benefit from services based on this data.
However, where public welfare is concerned, such as the safety of road users, there may be a need
for legislation on data access that needs to be followed by private actors. One example is with
emergency calls, where vehicle manufacturers need to provide access to data concerning accidents
via telecom operators to emergency services.
A similar situation concerning safety is now emerging in connected vehicles. In Europe, legislation
exists [38] that identifies categories where private actors need to provide access to data for free. This
includes categories such as the identification of a slippery road surface or the identification of wrong-
way drivers. In such cases, private actors are obliged to provide access to their data, if it contributes
to the enhancement of road safety.
Data owned by individuals
In principle, it is the individual that must decide whether the data generated by their vehicles can be
forwarded and used. [39] However, usually the individual has only limited influence on their data –
when signing a contract (e.g. for a vehicle or for a mobile network operator), the rights on data usage
is also generally handed over to the manufacturer or service provider.
Alternatively, when considering connected vehicles where vehicles exchange data with vehicles or
infrastructure, it is foreseeable that the vehicle owner may decide whether to provide access to in-
vehicle generated data. This principle is followed in Europe [40], where the owner of the vehicle needs
to give their consent to the use of data and to whom it may be given, including the specific purpose
AUTOMATED VEHICLES 2021R03EN
49

for the use of data and hence for the identified service. It foreseeable that there will always be an
opt-out option for end customers and data subjects.
Based on this general principle it is important for road network operators to inform travellers about
the purpose for which their data will be used. The willingness of travellers to accept a reduced level
of privacy will likely increase if a clear reciprocal benefit is perceived. To make the road safer or to
improve efficiency might be good argument to convince drivers to provide access to their data. In
parallel the supporting services need to be exposed and made understandable for road users. For
other service categories (such as law enforcement), national legislation needs to be prepared as most
drivers would not willingly volunteer hand data over to road authorities for enforcement purposes.
Especially in LMICs, law enforcement strategies might have a significant safety impact. Therefore,
national legislation needs to be prepared to implement such services.
Cooperation models to give access to data
In general, private actors have concerns opening their data to others, mainly based on competition
issues. However, to enable success through the automation of parts of the mobility system,
cooperation and data exchange between all actors involved – public, private, and individual – is
essential.
Therefore, cooperation models need to be prepared, mainly concerning the access to data from
private actors. Such cooperation models need to ensure that data supplied for road safety purposes
by private actors is not used for commercial purposes by competitors. In addition, it needs to be
ensured that the data accessed is used for the specific purpose concerned (e.g. for road safety
information) and not for competing commercial services without contributing to the target concerned
[41].
Along with the general and legal principles of a cooperation model the technical framework needs
also to be prepared. Usually, data formats used and interface descriptions to access data are part of
the cooperation model. Ideally such cooperation is set up widely, involving a broad range of actors.
This is important to avoid convoluted and potentially contradictory agreements between the
individual actors involved.
6.2.2 Legal aspects on data
As discussed above, with regards to access to data, state and national legislation needs to be
supportive in enabling data exchange between all actors involved in the automated vehicle domain.
In addition to the data access elements, privacy and security issues also need to be considered.
Privacy
Personal information protection is one key principle when dealing with data. The PIARC technical
report on “Big Data for Road Network Operations” [21] provides an overview of international
examples of personal data regulation. In principle, a “privacy by design” approach needs to be
followed when dealing with data and services. This principle needs to be implemented using technical
as well as organisational measures. Where personal data (e.g. individual data collected by the vehicle)
is concerned, mechanisms need to be in place to anonymise the personal data before forwarding it to
any data-processing entity. Also, the privacy principles should be clearly documented and applied
within cooperation agreements between different actors
AUTOMATED VEHICLES 2021R03EN
50

Only if legal legislation allows (e.g. for enforcement issues), can the collection of private data be
permitted and the “privacy by design” approach be avoided.
Data protection
Data protection is a core element, when it comes to data exchange and data access with regards to
automated vehicles. Two different aspects with respect to data protection need to be considered.
Firstly, the data protection within the data holding and data processing entities, which may be public
or private entities or an automated vehicle. Secondly the communication link to exchange data
between entities needs to be secure.
For a public or private company data protection is a core principle, as discussed in the PIARC technical
report on “Big Data for Road Network Operations.” [21] When it comes to the protection of data
within the automated driving system within the vehicle, it gets more difficult. Holistic concepts for
cybersecurity to avoid any manipulations of data that enable or support automated driving functions
are still under development.
To ensure secure communication, first it needs to be recognised that only necessary data shall be
exchanged between different actors. In addition, for the exchange of this necessary data between
different actors, a proper encryption methodology needs to be in place to ensure that each interacting
actor can trust the exchanged data. This is necessary as a preventative measure against attacks and
misuse of data. Concerning the encryption of the communication link itself, the different strategies
are discussed in the PIARC technical report on “Connected Vehicles - Challenges and Opportunities for
Road Operators” [21]
6.2.3 Cost of data
Data and service provision by a road network operator based on the marginal cost principle, as
discussed earlier, must serve the main interests of the operator - improving safety and efficiency and
to reduce the environmental impact of mobility. Accordingly, it should be in the interest of the road
network operator to provide relevant data to serve the functions that contribute to positive impacts.
Even if a public entity decides to provide data under the open data principle for free, that does not
mean that this data has no cost (to the public entity). Taxpayer’s or road user’s money is usually used
to purchase or collect data needed to provide proper services into the automated vehicle.
When it comes to data owned by private entities, where data is part of their daily business, data access
is usually based on commercial models, where road network operators might purchase specific
datasets. This is the current state of the art situation for traffic management issues as well as traveller
information services provided by road network operators. Nevertheless, in an automated mobility
system all actors will need to work closely together to deliver the desired benefits with regards to
safety, efficiency, and environmental impact. From the road network operators’ perspective, the buy-
in of data on the marginal cost principle might be an option. But to use taxpayer’s money to purchase
data from private companies at current market prices can be difficult to justify, along with the
complication of the potential inclusion of personal data of individuals. In this regard, current policies
of road network operators around the world are highly fragmented and not adapted to the emerging
or future issues with big data in in relation to the automated mobility system.
AUTOMATED VEHICLES 2021R03EN
51

However, as stated in Section 6.2.1, there are currently legislative processes underway that would
ease access to in-vehicle-generated data by RNOs. Access to safety relevant data is the starting point
with the potential for access to traffic flow data to follow.
When looking at future functions of automated driving systems, even emission data of individual
vehicles may be of interest to road network operators. All over the world “low emission zones” are
being introduced in cities and other sensitive areas. If an automated car can indicate it can operate
with low emissions, it should be permitted to enter “low emission zones”. Future policies might focus
on giving physical mobility access to specific areas based on reciprocal access to relevant vehicle or
other road user data.
When discussing costs of data, individual user needs are required to be considered. Will a future
traveller accept additional costs for data exchange? Most likely not, as travellers have a limited
willingness to pay for vehicle connectivity features. However, all actors including vehicle
manufacturers and telecom operators will have to look for ways to make services more affordable and
attractive to increase the willingness to pay.
6.2.4 Trusted data environments
Regarding access and data protection, the setup of a trusted data environment is crucial. All actors are
keen to trust data that is used for supporting automated driving functions. However, trust becomes
even more difficult to obtain when more data and service providers are participating. A key aspect for
the further development of automated functions within vehicles is how to ensure that data
transmitted by vehicles from different brands can be trusted. Similarly, how to ensure that data from
different authorities can be trusted – such as, motorway operators, regional road authorities and
urban road operators. The set-up of a trusted environment goes far beyond the simple data protection
principles discussed in Section 6.2.2 and needs to be based on the cooperation models discussed in
Section 6.2.1.
There are two main ways of setting up a trusted environment. Either it can be set up as central
infrastructure with a neutral server for data-sharing purposes, or it needs to become a decentralised
system, e.g. as cloud-based infrastructure, where access to data is based on common agreements.
When setting up a central server infrastructure, a trusted party will get access to all relevant data via
defined interfaces. Based on cooperation models the trusted party will provide access to the data for
single actors with required access and privacy controls. This will enable all actors to have access to the
agreed data for preparing services for automated in-vehicle functions. Such a model ensures that all
data is centrally stored, but the question remains “who is in the position to operate the central data-
sharing infrastructure?”
In a decentralised system a central entity needs to curate a catalogue-service providing metadata
information of accessible datasets including contractual issues for data access, such as costs of data
or appropriate usage of data. The data transfer afterwards is directly between different entities based
on the agreed cooperation model. Although the second model looks more complicated, from a
commercial perspective it currently looks more promising.
AUTOMATED VEHICLES 2021R03EN
52

7 IMPACTS ON ROAD NETWORK OPERATIONS (RNO)

7.1 IMPACT OF AUTOMATION ON ROAD NETWORK OPERATIONS
Automated vehicles and various automated functions, associated with human driven and connected
vehicles, are currently emerging and are yet to mature or be in widespread use. Although many
jurisdictions are making it possible for vehicles with varying levels of automation to use public roads,
the number of automated or partially automated vehicles operating on public roads is still very low.
Currently, there are no commercially available level 4 or level 5 automated vehicles. Most vehicles
with high levels of automation are the subject of trials and pilot projects.
In such low numbers and with technology and automation still in the development and testing phases,
the assessment of the impact of automation on Road Network Operations (RNO) cannot currently be
measured across most intended applications. Some pilot projects are assessing specific automated
functions and are able to indicate some early impacts in very narrow operating contexts. The current
expectations of impacts on RNO tend to be largely based on various forms of modelling or prediction.
Many predictions over the last decade or more have been publicised that claim wider adoption of
vehicle automation will result in significant positive change. More recently it can be observed that
prior predictions about large positive changes are being revised in some areas with more tempered,
and possibly more realistic, predictions as new research and understanding is developing with
increased real-world exposure to complex vehicle interactions.
Within this environment of changing expectations, many RNO are looking to understand their
contribution towards supporting and enabling the on-road operation of automated vehicles. Broad
impacts are expected on how roads will operate and be operated in an automated future. Adoption
of automated vehicles may impact how and where RNO invest in roads and road related infrastructure.
Changes in how vehicles interact and how roads are used could impact the existing traffic control
systems and tools and how they are used by RNO for both recurrent operating conditions and
disrupted conditions.
Two major areas of focus are the likely impact on roadway capacity or congestion and impacts on road
safety. A third major area of focus is the changed efficiency of vehicle travel (e.g. reduced fuel costs
and other associated operating costs through aspects of platooning and other forms of string
formation), especially for certain vehicle classes such as trucks involved in freight transport. Emerging
use cases involving automated shuttles and the ongoing shift towards shared mobility and mobility on
demand (mobility as a service - MaaS) also have the potential to impact RNO in diverse ways. The
potential impact on Road Safety is discussed in more detail in Chapter 9.

7.2 IMPACTS ON TRAFFIC OPERATIONS AND ROAD CAPACITY

One of the key areas of focus and potential benefits from AV on RNO is the increase in capacity of
existing road infrastructure.
There are several driver and vehicle interactions which in part determine the amount of traffic that
can be serviced, per unit of time. For example, there is a general expectation that Connected and
Automated Vehicles (CAV) will be capable of travelling safely with much smaller headways than
existing human controlled vehicles. Smaller headways lead to the broad assumptions that more cars,
AUTOMATED VEHICLES 2021R03EN
53

travelling closer together per unit of time, will be able to pass through a section of road, thereby
increasing the outflow and theoretically increasing the operational capacity of a road segment.
A second aspect that needs to be considered for any technological impact to be realised is the level of
penetration of the technology within the broader vehicle fleet and the available services that drivers
wish to subscribe to and utilise. For some automated functions, penetration rates may not need to be
high to achieve a reasonable shift in overall outcomes, however, other functions may require quite
high or virtually 100% penetration before substantial milestones can be achieved.
The combination of mature and proven automated functions that can truly impact capacity (positively)
and minimum penetration thresholds (e.g. greater than 40% for some functions) need to be met
before real impacts can be measured on public road networks. In the absence of these two conditions,
transport modelling (of varying levels of detail) is currently the primary tool(s) utilised to attempt to
understand a wide range of “What-If” scenarios.
Some recent studies have investigated the impact of changed vehicle operations in microsimulation,
which enables the individual vehicle interactions to be accounted for in the assessment. Vehicles are
modelled with various “driver” behaviour assumptions applied which control how they could interact
with other modelled vehicles, simulating the inclusion of automated vehicle functions. Used in
combination with reasonable estimates of penetration within the overall vehicle fleet, some useful
and enlightening results have been published which consider the potential changes to operational
road capacity over time with increased CAV penetration and technology changes.
For example, the Karlsruhe Institute of Technology, Institute for Transport Studies along with the
Technical University of Munich in partnership with PTV Group undertook a detailed study, focussing
on the impact of automated vehicles on uninterrupted flow on German Autobahns. The work was
undertaken on behalf of and with the cooperation of the German Research Association of Automotive
Technology (FAT) [42].

The outcomes indicate that rather than there being a significant expected increase in the operating
capacity of German autobahns. It was found that the likely outcomes were less optimistic than many
previous studies. Of primary impact were factors of slow CAV penetration and restrictions on the
ability of automation functions to truly push the limits and take “risks” beyond what human drivers
would currently consider.

In the near to medium term (up to 2050), the study indicated operating capacity is expected to
decrease. This is due to low penetration rates combined with limitations on automation where
automated vehicles would be required to interact in a more conservative manner than human drivers.
This more conservative behaviour is due to current limits in sensors, ongoing maturing of
communications technologies and systems and to some extent exposure to the risk of developing
technology not functioning, not reacting as expected or not as quickly as required.

In the longer term (beyond 2050), modest increases in average operating capacity are forecast,
however, significant changes in connected and controlling technologies are required as well as
significant levels of penetration of vehicles with level 4 / 5 automation. Even under such scenarios,
there is recognition from the study that different types of road segments will be subject to varying
levels of operational capacity increases. Extension of the modelling outcomes across the whole
AUTOMATED VEHICLES 2021R03EN
54

autobahn network across Germany in the study suggests that is may be possible to achieve a 30%
increase in capacity and therefore a reduction in traffic delays and user costs. It is noted that results
derived under a similar approach to modelling traffic in in motorway networks in other countries may
show varying outcomes due to the unique nature of motorway segment types and the extent to which
they may impact on overall network capacity.

Figure 42 demonstrates the expected capacity changes across different types of motorway segment
types. The segment types assessed were basic, merge, diverge and weave sections and are consistent
with the segment definitions within the German HBS. It is also acknowledged that in road networks,
isolated capacity restraints can restrict traffic from accessing or utilising downstream and upstream
portions of the network, either by restricting flow passed a bottleneck or causing congestion and
queueing on the approach. Figure 43 demonstrates the expected decrease in capacity on German
autobahns with increases in capacity above current day not expected until beyond 2050.

Figure 42: Simulation results representing the percentage change in capacity of each freeway
segment relative to the base scenario
AUTOMATED VEHICLES 2021R03EN
55

Figure 43: Impacts on AVs on the freeway capacity in Germany

The results from this study may seem conservative, however, they are considered plausible based on
the assumptions applied in the analysis. The procedures take into account a mixture of more complex
vehicle interactions beyond simply reduced headways. The simulation accounts for lane changing and
merging movements which require minimum necessary gaps between vehicles, this imposes a
limitation on how small headways can be practically. Significant rates of lane changing have been
identified as being necessary for freeways to operate at capacity flows, especially on multi-lane
facilities in urbanised areas. Evidence is also emerging that existing vehicle headways in near capacity
flows are lower than what has been used for previous capacity assumptions using average headways.
Automation of lane changing in highly complex, dynamic and chaotic conditions also requires
significant future development to limit disruptive vehicle interactions and ensure safe operations.
The German autobahn study is limited to freeway operating conditions, which is expected to be the
part of the road network that is most amenable to early automated vehicle operations. Varying
operating conditions on other parts of road networks are likely to result in different impacts on
operational capacity. In a similar manner to the freeway environment, care should be taken that a
generalised increase in capacity is not necessarily assumed. The many complex interactions that take
place on the road network, such as interactions with other road users (pedestrians and cyclists) could
make full automation and reliance on automated/semi-automated assistance difficult to implement
efficiently and safely.
In addition to freeways, similar assessments and monitoring are required for arterial streets where
different interactions take place, especially with a greater variety of different transport modes,
including active modes (walking and cycling). Automated vehicles are “defensive” and are
programmed to stop or take evasive action when “obstacles” are detected. The arterial street
environment presents many more of these potential situations that can trigger such defensive
responses and lead to disruptive events and reduced capacity.
The potential for such disruptive events has prompted considerations to segregate or dedicate space
(routes) to automated vehicles. If segregation were provided, impacts may still be experienced at
transition zones and the disruption may be concentrated near the desired segregation point(s).
AUTOMATED VEHICLES 2021R03EN
56

Some detailed microsimulation studies of isolated aspects of arterial street operations, such as
increased traffic flow through traffic signals during green windows, has been undertaken. Results need
to be considered with regard to the variety of interactions that may occur in real world situations
which may not have been incorporated into modelling considerations. The impacts of upstream and
downstream capacity constraints in the network (at origins and/or destinations) may also impact
outcomes and may not be considered in models with limited extents. Interaction with other road user
classes tends to be more varied in the real world in comparison to modelling scenarios and is likely to
significantly impact arterial street outcomes. Although benefits may emerge for arterial streets over
time, they may be reduced compared to ideal operating conditions where greater complexity exists.
RNO need to be realistic about the potential changes in road capacity. While the longer-term impact
on capacity may be positive, there is emerging acknowledgement that the road operating
environment is likely to go through a transition period where capacity is likely to be reduced at first.
Some suggestions have been put forward that road agencies may be able to invest less in expanded
road infrastructure over time if operational capacity increases, although increased demand could
increase maintenance costs due to accelerated pavement fatigue. However, if the likely scenario is a
reduction in operational capacity in the near term, care is needed to ensure adequate physical
capacity is maintained and/or appropriate improvements are still considered for the near to medium
term. Emerging understanding would indicate that reducing physical capacity in the expectation that
CAVs will increase operational capacity would be unwise.
There is currently limited real-world indication of increasing automation inducing further vehicular
demands on road networks, due to the limited penetration of automated vehicles. There are
expectations that automaton may result in more vehicles circulating on the road network, some with
no occupants either returning to an origin or transiting to service an on-demand mobility request.
Whether capacity reduces, remains relatively fixed or increases, such increases in on-road demand
would inevitably impact road network capacity and operations, especially in urban environments.

7.3 PLATOONING
Platooning is the ability for multiple connected and automated vehicles to travel together at the same
speed and with short headways, following each other in a concentrated group or platoon. Connectivity
between vehicles is required for platooning to function efficiently and safely. A direct V2V (vehicle-to-
vehicle) communications link is considered necessary for platooning to function appropriately. In
addition, V2I connectivity may also be desirable to enable monitoring of platooning activities and
allow for potential optimisation operations (e.g. activation of signal operations to assist platoon
progression).
Several benefits have been identified that platooning operations could deliver:
• Energy savings and reduced fuel consumption through reduction in traffic oscillations and also
aerodynamic drag, especially for trucks and heavy goods vehicles,
• Emissions reductions through reduced fuel consumption
• Improved road safety by applying effectively simultaneous braking through V2V
communication and reduced speed / traffic oscillations
• Some potential capacity efficiency improvements are considered possible since generally
more vehicles could operate closely spaced (increased density).
AUTOMATED VEHICLES 2021R03EN
57

Over the last three decades, many pilot projects and experiments have been conducted to test and
prove the necessary technologies for platooning, being a fusion of many existing automated functions.
In some more recent cases, impacts of platooning have been assessed, although on a relatively limited
sample.
Measures from the impacts of platooning have primarily been limited to energy savings indicated by
reduced fuel consumption. Results from early pilot projects have indicated a 10-17% improvement in
these measures, although it was noted that a number of factors may reduce the ability to achieve
these benefits, such as interactions with conventional (non-automated) vehicles, road geometry and
vehicles configurations. In addition to the efficiency improvements, platooning operations may have
the potential to improve road safety and operational capacity, although these are currently difficult
or not possible to measure given the limited scope of platooning studies and pilots, and their limited
operating environments.
Planning and assessment for platooning operations have highlighted some challenges:

• Length of platoon is bound by V2V speed and reliability to maintain string stability
• Length of platoon must be managed to avoid bottlenecks at highway/freeway entrances and
exits
• Theoretically the longer the platoon, the greater the fuel savings, therefore, shorter platoons
may limit efficiency benefits
• Practically, long platoons can be disruptive to necessary merging and weaving of non-platoon
traffic
• Limitation on the ability for a platoon to change lanes if required and over significant lengths
(such as lane reductions on freeways)
• In mixed traffic, cut ins by non-platoon traffic present the biggest challenge to maintaining
platooning integrity
The need for limited interruption of platoons and a desire for longer rather than shorter platoons has
led to consideration of where best to utilise platooning within the road network. Urban and rural
freeways may need to be further investigated. Urban freeways for example typically have close
interchange spacing, short trips and high flows, resulting in a greater “turnover” of vehicles in the
roadway than on rural freeways. High turnover can impact platooning operations, even if the entering
or leaving vehicles are not part of an operating platoon, due to the need to interact with, pass-through
or pass-by.
To limit disruption, dedicated lanes could be considered to enhance platooning safety since the
behaviour of platooned vehicles would be more predictable and speed would be more consistent;
however, understanding the impacts for traffic accessing / leaving dedicated lanes would be
necessary. Disruption caused by manoeuvring into and out of “platooning lanes” could also reduce
the overall benefits of the platooning operations. Early operations could also result in underutilisation
of dedicated lanes where provided, potentially requiring a transition period with under-utilisation of
physical road space.
Platooning may not be feasible or appropriate in areas with significant variation in road grade as
aggressive speed adjustments may be needed to maintain platoon integrity, reducing the ability to
meet desired operating efficiencies. Intelligent monitoring and control systems could account for
AUTOMATED VEHICLES 2021R03EN
58

topography and optimize fuel efficiency while traveling over hilly terrains. Platooning control systems
would also need to consider the combined response to unexpected events such as vehicle breakdown,
tyre punctures or wildlife strikes.
Platoons of heavy goods vehicles may have impacts on road assets through increased wear and tear
on road pavements. Structural implications also need to be assessed and monitored due to closely
spaced live loads. Increased resources for monitoring, maintenance and deploying preventative
measures will likely put additional loads on road network operators and asset management agencies.
Further investigation is also required to determine the need for supporting road infrastructure that
needs to be provisioned and maintained by road network operator.
Further assessment is required to identify the use cases for platooning that are able to provide the
best outcomes for road users, such as heavy goods vehicles. Platooning functionality may only be
possible on a limited number for public roads, such as freeways, and even then, certain conditions
may need to pre-exist for reasonable levels of platooning to be considered. Another key aspect is the
need for a lead driver (or platoon controller). This may need to be a skill considered essential for
platoon formation which poses issues when a platoon controller (and associated vehicle) no longer
need or desire to remain with a formed platoon. This may imply that all drivers have minimum skills
and training to enable platoon control on as as-needs basis, which a road operator may have some
involvement in setting requirements.
Road operators also have a role in influencing and changing regulations that may enable platooning
within their jurisdictions. The framework for considering such changes would need to consider both
the real benefits that can be achieved as well as the operational risks associated with complex
conditions that exist on traffic network. It is noted that some leading truck manufacturers have
recently updated their assessment of the economic benefits of platooning which may impact the real
benefits that could be expected across logistics fleets.

7.4 AUTOMATED SHUTTLE MUST BE CONSIDERED IN AN URBAN AND RURAL ENVIRONMENT

The deployment of digital technology and new technologies enables outcomes that were previously
unachievable. It provides access to new services - carpooling, car sharing, transport on demand, or
even bikes and scooters. It promotes the organization of connections between modes and gives very
broad access to all mobility services.
7.4.1 Automated public transport systems, an opportunity to develop new services
The rapid development of automated public transport systems positions them as a future component
of the public or private transport services offering. Some are marginal yet present in various
jurisdictions, particularly in urban areas, and with prospects for further application in peri-urban or
rural areas. Although business models may be yet to mature, it is reasonable to consider the prospect
of increases in such services in the future. Automated public transport systems have the potential to
provide solutions for unsatisfied mobility demands alongside existing public transport services, such
as transportation in areas with low density (rural), low flows or demand for first and last kilometre
trips.
AUTOMATED VEHICLES 2021R03EN
59

7.4.2 Services for both urban and rural areas

In rural areas, improving the availability of transportation is key. Rural areas are often very poorly
serviced by public transport systems and can struggle to provide adequate public transport for their
residents. Continuation of this lack of access to mobility services can increase the risk of isolation of
the affected population. Automated public transport systems are a potential solution to address this
need, especially for elderly people who are no longer able to drive. In Japan, where more than 25% of
the current population are 65 years old or older, several autonomous driving deployment projects
initiated by the Japanese government in rural areas are specifically targeted to cater for elderly
people. Following field operation tests (FOT) by the Cross-Ministerial Strategic Innovation Promotion
Program (SIP), shuttles have been in operation with fares collected by a local NPO since November
2019.

Figure 44:Automated driving of shuttle in rural area (Kamikoani village, Japan)

According to the United Nations [43], in the next 20 years, 60% of the world's population will live in
urban cities. This will lead to a large increase in urban travel demand which can be facilitated by
automated public transport systems. In many urban cities, reducing the use of individual cars is a key
objective. Many cities have adopted such a mode shift (from the use of individual cars to public
transportation) as central pillar of their transportation policy. Such a change requires the promotion
of alternative transport solutions that encourage drivers to reduce or cease private vehicle utilisation:
automated shuttles are seen as one of the key future alternatives.
Foreseen use-cases for public / shared transport by the French Ministry for the Ecological and Inclusive
Transition (in charge of Transportation) include:
• Access/egress from stations for the first and last kilometres in sparsely populated areas
• Service in a limited territory, with pre-defined stopping points (e.g. commercial zone, hospital,
university, tourist centre)
• Night services on existing lanes by enhancing geographical scope and frequency
• Capacity adaptation (modular platooning) (passenger transportation capacity that adapts to
demand by modulating the number of shuttles in platoon mode according to demand)
• Integrated shared mobility + parking services
• Shared mobility on HOVs in peri urban zones
AUTOMATED VEHICLES 2021R03EN
60

7.5 HOW RNO CAN HELP CAV ON WORK ZONES/INCIDENTS

The sensors onboard automated vehicle are a key component for analysing the road and surrounding
operating environment. If onboard sensors are unable to operate effectively, an onboard digital
mapping system is needed to supplement the missing information. Road works or incidents may also
alter the road layout, changing where the vehicles are expected to travel. The challenge for CAVs is
that the scene around an incident may be difficult to interpret. The normal rules of the road may give
way to a common-sense approach in which vehicles navigate slowly around unusual obstructions, such
as disabled vehicles or debris on the road.
Drivers may be required to interpret manual signals from police officers or other traffic management
personnel. Whilst live incident reporting may offer some details of the nature of the incident, the exact
configuration of any traffic management measures on the approach to the incident may not be known
until the CAV arrives at the incident location and may also be subject to change over time.
A solution to address these challenges is to transfer control of the automated vehicle back to the
driver, however this may not be ideal if the driver is not prepared to take control and the situation
could deteriorate. Another solution is for the road network operator, through connectivity, to provide
I2V services aimed at conveying reliable, accurate, and comprehensive information on Road works
(eg: RWW- Road Works Warning – see below) or Incidents to automated vehicles. From a road and
traffic operator’s perspective, communicating this type of information to CAVs can also improve road
worker safety, for whom work zones can be dangerous or even deadly.
Road Works Warning (RWW)
The general concept of this service is to enable a RNO to communicate with drivers through
communication from the infrastructure to vehicles (I2V) about road works, restrictions, and changed
instructions. The service provides information to approaching vehicles about conditions at a work zone
ahead such as the work zone characteristics and activities that could present unsafe conditions. This
includes information such as obstructions in the lane or lanes ahead, lane closures, lane shifts and
speed reductions. The roadworks site can be fixed or mobile, planned or unplanned. In either case,
drivers are warned about the state of impacted lanes and presence of vulnerable workers. Examples
of projects using Road Works Warning include SCOOP@F (FR) , US DOT data exchange project (USA),
Eco-AT project (AUS - GE –NL), C-ITS Pilot Project (South Korea)
Regarding immediate response/emergency situations, a “CAV Compliant First Respondent’s
Procedure” could be developed. Such a procedure outlines a series of steps that the first authority on
scene should follow. A suitable early step may be to set-up warning signs / signals / mobile roadside
units (RSU) which alerts nearby CAVs that the area is in an unusual state and that it should either be
avoided or navigated carefully. Signs / signals for this purpose should be easily identifiable by CAVs.
Close corporation with organisations developing CAV technology would be required in developing
such procedures. Report Summary - Transport Systems Catapult – 2017

7.6 TRAFFIC MANAGEMENT BY RNO

As discussed earlier, sensors onboard automated vehicles analyse the road and surrounding operating
environment. If the analysis is insufficient, an onboard digital mapping system is needed that can
AUTOMATED VEHICLES 2021R03EN
61

supplement the missing information. Generally, an automated vehicle’s ability to perceive its
environment is based on data fused from embedded sensors, GNSS/mapping navigation and V2X.
The information about the surrounding physical infrastructure will therefore become increasingly
digital. The challenge then is to build this digital infrastructure and to keep it up to date as conditions
change. CAVs require up to date information to enable the selection of the best route, and also ensure
that they position themselves appropriately and safely in the roadway. This information needs to
include roadworks, road closures and diversions, including detail about the location, dates of
disruption and scale of disruption. This digital infrastructure also raises the issue of data exchange
standards and formats.
In Australia, Austroads published a Connected and Automated Vehicles (CAV) Open Data
Recommendations report on 23 August 2018 [24]. According to this report, there is a need for a
consistent messaging framework that would allow RNOs to query the required data from vehicles.
RNO data types that were identified to be a high priority for CAVs include:
a. Live feeds from traffic management systems for variable speed limits and lane
closures
b. Live feeds of traffic signal phase and timing data (such as SPaT messaging)
c. Available data for emergency road closures (fire, flood, etc)
d. Available data for temporary conditions associated with works, events, and incidents
e. Advance notification of new and changed roads (that may not have been mapped)
f. Coordinate with actions already occurring on speed limit data and extend to cover
other traffic restrictions (vehicle size and mass)
Changes to business processes may be required for RNO to supply the most suitable data to CAVs. An
example is the need for RNO to be able to supply the most relevant and useful data about roadworks,
events and incidents on the main road networks, which may require a higher level of accuracy and
timeliness suitable for the ODD which may not have been the case in a non-CAV scenario.
AUTOMATED VEHICLES 2021R03EN
62

8 RESPONSIBILITY AND FINANCING

8.1 RESPONSIBILITY AND INSURANCE
There are a number of challenges and unknowns with regard the responsibilities and obligations of
the various actors involved in CAV deployment and operations, including:
• Responsibilities and obligations of drivers and owners: who or what determines whether the
driver of an automated vehicle must be vigilant? Are existing rules of civil liability sufficiently
clear for insurance companies?
• Responsibilities and obligations of car manufacturers: under what conditions can a car
manufacturer test and market its vehicles? Are any extant deviations from existing rules of
penal and civil responsibility demonstrably justifiable?
• Responsibilities and obligations of RNO: what should they provide and to what level of
service? What happens if that level of service is not granted?
The following sections provide some aspects for RNO to consider.
8.1.1 Insurance and responsibility of the driver
In France, since 1985, there is a law ensuring a fast-financial compensation for victims of car crashes
(“Badinter” law). The victim can claim compensation from the driver or, if he/she is not identified, to
the owner of the vehicle, who has an obligation to have an insurance covering civil responsibility.
Therefore, from a civil responsibility point of view, the case of an automated vehicle does not change
anything. Insurance of the owner, if not the driver, will provide compensation to the victims.
Insurance costs of an automated car are as uncertain as the road safety benefits. It is probable that
the first automated cars will have higher insurance costs due to some recent accidents in the USA
involving automated operations, although this would be expected to decrease once liability of the
system is widely admitted.
Penal responsibility, however, will depend on the instructions given to the driver, and will still be upon
the driver for lower SAE levels, if the driver is supposed to remain alert to the road conditions and to
react when required.
8.1.2 Responsibility of the car manufacturer or of the car
If an accident occurs when the automated mode is active and the vehicle has not asked the driver to
regain control, the responsibility may apply to the car manufacturer. This responsibility would be
linked to the homologation of the vehicle and its conformance with the homologation framework.
This framework will probably include the ability to update, as systems that are introduced in the next
few years are unlikely to be perfect nor remain in place for extended periods. As versions become
outdated, it is likely that many earlier conforming vehicles will still be operating on the road. A key
goal should be to limit the physical risk of these systems through a variety of technical and contractual
tools to enable monitoring, over-the-air updates, and even virtual recalls as necessary.
Some lawyers consider that automated cars and more generally robots should be considered legal
entities, and that a new brand of laws should be dedicated to them, building upon the “3 laws of
robotics” of the science-fiction writer I. Asimov:
AUTOMATED VEHICLES 2021R03EN
63

1. A robot may not injure a human being or, through inaction, allow a human being to come to
harm.
2. A robot must obey orders given by human beings except where such orders would conflict
with the First Law.
3. A robot must protect its own existence if such protection does not conflict with the First or
Second Law.
8.1.3 Responsibility of the RNO
In France, the responsibility of the RNO is to perform “normal maintenance”. If this is not the case,
the RNO may be considered to have a responsibility if an accident occurs.
The question is how the concept of “normal maintenance” will evolve with automated cars. It is
reasonable to expect that a minimum level of service is defined for the contrast of road markings to
support CAV operations for example. However, it is technically impossible to predict or detect the
exact time when the contrast will fall below a given intervention threshold and to then enact
remediation instantaneously. Such deterioration below a threshold is a random outcome related to
the number of wheels having crossed the line, the weather conditions over time and the quality of the
asset when provided. This implies that vehicles will always need to cope with situations with damaged
markings (potentially with the assistance of a reconstructing pavement marking algorithm or
information from a HD map) and that “normal maintenance” may rather be defined by a frequency of
surveillance of the markings. A similar approach needs to apply to connectivity, which cannot be
guaranteed to work 100% of the time. This tends to be well integrated by car manufacturers to date,
with connectivity considered as an additional sensor, thereby providing redundancy and including
fallback solutions based on the vehicle’s optical and other sensors, HD maps and consideration of the
ODD.
Some innovative automation use cases rely on the infrastructure to provide a part of the artificial
intelligence (enhanced perception through sensors on the infrastructure, guidance through waypoints
in roadworks zones, or even supervision of specific fleets such as public transport or platoons). This
implies more responsibility by the road operator and/or the road equipment provider to both deploy
the additional supporting physical and digital infrastructure and provide the information to CAVs.

8.2 COSTS, BENEFITS AND FINANCING

8.2.1 Benefits expected
In general, most of the research today, indicates uncertainty about the possible system effects of the
ongoing technical and digital development on the road transport system. From a business perspective
however, it is clearer that automation, electrification, and digitization in the short and medium term
will result in significant benefits. When it comes to travellers, the picture is more fragmented, as their
choices are not only governed by economic rationality, but also by individual values, lifestyles, and
acceptance of new services.
The development of the new technology, regardless of traffic type, has the potential to improve the
accessibility to the road transport system, capacity utilization, increased traffic safety, and reduced
climate emissions. These benefits, however require well thought out implementation of the new
services and functions in societies with minimum negative side effects (such as; pricing, cyber security,
AUTOMATED VEHICLES 2021R03EN
64

counteraction of digital segregations, etc.), and sufficient incentives that promote behavioural
changes (circular and shared economy).
8.2.2 Estimation of costs
It is quite difficult to advise on the investments needed for digital transformation of road transport
systems as the size of the gap between the investor communities and the digital industry varies heavily
from one country to another. The challenge for the public sector in this context is its low investment
commitments despite the recognition of the importance of the digital transformation for developing
a safer and efficient road transport system. However, the gap between the current status and the
desired digitalization needs to be studied, required investments estimated and carefully planned and
funding institutions (both private and public) identified in order to implement effective of measures
to drive the development forward. The plan needs to also take digital integration into consideration
as most capital investments today tend to be focused on urban areas. Developing initiatives and
financing solutions for the digitalization projects in rural areas is an important aspect in this context.
A recent study estimates the cost of the investment needed in IT communication infrastructure to
enable large scale deployment of V2I as EUR5 - 8bn in the EU and US$7-12bn in the US. This study was
conducted by Ricardo Energy & Environment on behalf of 5G Automotive Association and submitted
in August 2020 [44]. For the achievement of this scale of investment the report urges broad
cooperation between cities, road operators and mobile network operators.
AUTOMATED VEHICLES 2021R03EN
65

9 SOCIAL ISSUES AND THE CONSIDERATION OF HUMAN

FACTORS
9.1 MAIN PRINCIPAL QUESTIONS FROM THE PERSPECTIVE OF SOCIETY
9.1.1 Motivation for seeking automation
At the outset, a fundamental question that is rarely fully articulated but should be asked is “why
automate in the first place?” [45] According to [46] there are a range of potential motivations for
seeking to automate human functions including:

• When it is dangerous for humans to carry out a task (e.g. driving impaired).
• When it is impossible for humans to carry out a task (e.g. accurate night-time sensing or
sensing beyond line of sight).
• When carrying out a task is difficult for humans (e.g. rapidly reacting to sudden obstacles).
• Just for the sake of automation (e.g. as a way of creating a market even if no safety benefit is
conferred to users).
The current development of automated driving systems is motivated by all of these in different
measure.
Perhaps a more safety-relevant way of addressing the issue of what motivates automation centres
around the two questions, “What tasks can automated systems do better than humans (from a safety
perspective)” and “what tasks can humans perform better than automated systems?” These are
central questions that have accompanied many past technology developments and articulated by Fitts
[47] in reference to machines in general within in a framework that has remained surprisingly relevant
today, even in the context of vehicle automation and technologies that were barely imaginable at the
time [48], [49].
9.1.2 Potential safety benefits
Foremost among the predicted benefits of increased and complete vehicle automation is the promise
of improved safety. With few exceptions [50], [45] this predicted benefit is accepted uncritically based
on the observed rate of human error-involved crashes. Claims of a more than 90% reductions in road
traffic deaths resulting from automation eliminating crashes linked to human error are untested. It
seems likely that the number of road casualties will decrease with automation, but crashes will not
disappear. In certain circumstances, more crashes may occur among “average” drivers that are not
prone to risky behaviour. This is particularly likely in circumstances where drivers are required to take
over from automated driving in emergency situations.
Much of the focus on the potential safety benefits of automated driving has been centred on the
elimination of human error in the driving task. That is because humans make mistakes in judgement,
may drive impaired or distracted, may simply not be adequately aware of the driving environment or
may not react quick enough to rapid or unexpected changes. The Safe System approach inherently
recognizes the capabilities and limitations of humans when designing and operating road transport
systems.
Errors arising from interaction with the traffic and road environment can be limited by understanding
these interactions and designing the road transport system from these interactions, in order to guide
AUTOMATED VEHICLES 2021R03EN
66

the road user to behave in a way that is as safe as possible. Yet, as human error cannot be fully
eradicated there is a need, at the same time, to mitigate the consequences of mistakes. In simple
terms, this basic principle of a Safe System starts with the insight that human error should no longer
be the primary cause of crashes. Instead, road crashes are a consequence of latent failures created by
decisions and actions within the broader organisational, social, or political system which establishes
the context in which road users act.
The lack of experience and data complicates an assessment of how safe automated driving really is. It
is further complicated by the lack of a common framework for such a safety performance assessment
and by rapid changes in its object; a self-driving car is a combined hardware and software system
whose critical performance characteristics can change radically with a software upgrade.
Vehicle automation strategies that keep humans involved in the driving task seem risky. A shared
responsibility for driving among both automated systems and humans may render decision making
more complex rather than simpler. Therefore, the risk of unintended consequences that would make
driving less safe, not more, could increase.
Humans retain an advantage over single sensor-based automated systems in many contexts.
Overcoming this gap requires combining input from several sensors. In some cases, safe operation will
require vehicles to communicate with each other and with infrastructure beyond the line of sight.
However, relying on this connectivity for safety performance involves risks, especially regarding
cybersecurity. Whether vehicle automation should move from a reactive safety paradigm (where
vehicles rely on their own capabilities) to a proactive safety framework (where vehicles are embedded
in a communicative network) is still debated. Potential uses vary in different countries according to
GDP, population density, technology advancements, etc.

9.2 THE LIMITATION OF POTENTIAL BENEFITS OF AV BY A RANGE OF HUMAN FACTORS

It is possible that the potential safety benefits of C-ITS and AVs may be undermined by a range of
human factors if not properly addressed. C-ITS and automation will change the driving task, either by
providing additional sources of information or by automating some, to all aspects of the driving task.
This can lead to changes in driver behaviour, some of which may be unintended by the system
designers. In road safety, the term ‘behavioural adaptation’ (see e.g. [51]) is typically used to refer to
the unexpected or unintended behavioural changes that appear in response to the introduction of a
change in the vehicle or road environment, that may influence the predicted safety benefits of this
change. The anticipated behavioural adaptation factors leading to changes in driver behaviour are not
always anticipated by vehicle and system designers. Some of the human factors included:
• Technology over-reliance, causing issues when drivers are required to regain vehicle control
or use a non-equipped vehicle.
• Driver overload (from system status monitoring), or underload or loss of vigilance potentially
leading to reduced situational awareness and difficulties coping with sudden demand
increases, such as during resumption of manual control. Issues with loss of vigilance and
situation awareness may be particularly apparent with SAE Level 3 or ‘conditional
automation’.
• Driver distraction when startled by alerts, having their attention drawn away from critical
information or when engaging in distracting activities while in charge of an automated vehicle.
AUTOMATED VEHICLES 2021R03EN
67

• Drivers failing to trust and/or accept the technology, leading to system misuse or disuse.
• Loss of driver skill, leading to problems in the event of automation failure and resumption of
manual control.

9.3 NEW HUMAN FACTORS CAN ARISE

9.3.1 Driver Overreliance (Automation Complacency)
Over-reliance, also called ‘automation complacency’ [52] or ‘delegation of responsibility’ is an
important human factor when introducing C-ITS and automation. Over-reliance occurs when drivers
delegate full responsibility of driving tasks to the system (regardless of whether the system is intended
to assume full responsibility or not), or delegate responsibility for other driving tasks that the system
was not designed to address. Over-reliance can occur because of a loss of vigilance or drivers
misunderstanding the functionality and limitations of the technology. This creates problems when the
system is no longer active, such as when drivers use a non-equipped vehicle or when drivers are
required to regain vehicle control. Over-reliance has been observed in several Intelligent Speed
Assistance (ISA) studies, where drivers, forget to change speed upon entering a different speed zone
when the ISA is no longer active [53]; [54]. Numerous instances of over-reliance on Adaptive Cruise
Control (ACC) systems have also been reported. For instance, drivers were observed to brake later in
response to a braking lead vehicle when using ACC [55]; [56] and some drivers failed to reclaim vehicle
control and collided with a lead vehicle when the ACC failed without warning [57]. However, drivers
can effectively assume control from ACC in critical braking situations if provided with a warning [58].
Educating drivers about the capabilities and limitations of C-ITS and AV technologies and providing
timely warnings when the system moves outside of its performance envelope can help reduce the risk
of overreliance.
9.3.2 Adoption of Risky Driving Behaviours
If C-ITS and AV technology are perceived as providing safety benefits, they can change drivers’
perception of driving risk and encourage risky driving behaviour. Drivers may adopt risky driving styles
to adjust their level of preferred risk, to experiment with the system, or to improve their mobility and
compensate for factors such as lost time, due to lower speeds created by some systems. When using
ISA, for example, drivers compensated for slower overall speeds by driving faster on roads where ISA
was not active, or in situations that warranted lower speed, such as when turning or in poor weather
[53]; [54]. The period of mixed traffic, where automated vehicles will share the road with non-
automated vehicles, may also present issues in terms of the behaviour of drivers of non-automated
vehicles. Issues may occur for example, if drivers of non-automated vehicles expect automated
vehicles to behave in the same (suboptimal) manner as non-automated vehicles and this expectation
is not met [59]. A preliminary analysis of the crash rates of automated vehicles, [50] found that
although automated vehicles had an almost five times higher crash rate than conventional vehicles,
none of the crashes were the fault of the automated vehicles. Such findings suggest that the drivers
of conventional vehicles may have uncertainty or difficulty reconciling how they would expect
automated vehicles to behave. Alternatively, drivers of non-automated vehicles may adopt similar
behaviours to automated vehicle platoons which are incompatible with safe manual driving. For
example, in a simulator study [60], participants driving non-automated vehicles were observed to
AUTOMATED VEHICLES 2021R03EN
68

adopt similar following behaviour to automated vehicle platoons, by driving at a reduced headway to
lead vehicles.
9.3.3 Loss of Skills
Automating parts of the driving task may lead to a loss of skill and this problem is likely to increase as
the level of automated driving increases [61]. If humans do not perform a task for a period, they begin
to lose the skill to perform that task effectively, even if they were able to perform it to a high standard
previously. Loss of skill can lead to problems in the event of automation failure where the driver is
required to regain manual control. This requires a skilled driver who needs to perform the manual
control tasks, as well as the automation, which often occurs under time pressure. A challenge for
system developers, manufacturers and governments will be designing automation and implementing
policies, to ensure that drivers can maintain a minimum level of driving skill (i.e. by requiring
intermittent manual control) or removing the need for drivers to intervene at all in the driving task.
9.3.4 Other issues
Driver Workload
An attempt to reduce workload through driver support systems can potentially increase driver mental
workload, as these systems add information which must be monitored by the drivers. Humans are
poor at monitoring tasks [62] and research has shown that monitoring systems is stressful and can
cause high levels of workload [63]; [64].
Driver Distraction
Automated systems can be a distraction risk, if they startle the driver with alerts, if they present
confusing, excessive, or false alerts, or they divert the driver’s attention away from safety-critical
events. We can also expect that there will be limited common standards for warnings and alerts
among different competing car manufacturers.
Driver Acceptance & Trust
Acceptance of AVs by drivers is a critical factor influencing the successful uptake of these technologies
and their effectiveness in improving road safety. A failure of drivers to accept a technology can lead
to them not using the system in the manner intended or failing to use it at all. Acceptance is closely
linked to driver trust – if drivers do not trust that a system is reliable, safe, secure, and effective, then
they are unlikely to find it acceptable. But reality shows that people accept new technology more than
it appears in attitudes surveys.
Regaining Manual Control
A whole host of factors including loss of skill, loss of situation awareness and overreliance can cause
issues with drivers regaining control of an automated vehicle. Reduced situation awareness, for
example, has been associated with a delay in appropriate braking when a failure in ACC was
encountered [65], [66]. Some research show that partial AVs will worsen distracted driving, by lulling
drivers into complacency. Tests at Virginia Tech found drivers took an average of 17 seconds to
respond to takeover requests from Level 3 AVs.
HMI Issues
AUTOMATED VEHICLES 2021R03EN
69

As more systems are introduced into vehicles, integration, particularly of aftermarket systems, will
become increasingly important, as drivers will need to be able to distinguish between information
from multiple systems, and the systems will need to communicate with each other to ensure that any
alerts or information provided are prioritised so as to not overload and confuse the driver with
multiple sources of information at once [67].
9.3.5 Examples of 11 new social issues and challenges that will arise
Challenge 1: Today, the work of industry and the discussions with national and EU stakeholders
(governments and their agencies) concentrate on research, testing and type approval. First activities
which are focussing on the development of traffic rules have started in some countries. What is the
total scope of affected policy and regulation?
Challenge 2: On type approval regulation: Level 2 is still under strong discussion, Level 3 has not really
started yet, and as for Level 4 and 5 there is no clear view on how to proceed. How can the type-
approval approach evolve? How to set up the regulation in a timely manner to ensure that it is in place
when the technology is ready? How to develop regulation and technology in parallel, in a harmonised
way, without creating a chicken-and-egg dilemma? How to deal with software updates?
Challenge 3: How and to what extent would it be required to adapt and harmonize traffic rules for a
quick introduction of higher automation levels?
Challenge 4: What liability framework needs to be in place to facilitate market penetration from a
legal/liability perspective?
Challenge 5: AVs will potentially grow total vehicle miles travelled (VMT) by 14% [68]. What policies
should be taken to motivate people to reduce car use or to share? To what extent should governments
limit freedom of travel? What kind of new traffic pricing policies should be taken?
Challenge 6: The transition period when roads will be shared by both conventional and automated
vehicles will probably be very challenging. It is unclear whether roads would be safer or more
dangerous during this transition period. Increased risk during the transition period is, in part, due to
the likelihood of drivers becoming over-reliant on technology and taking risks they might otherwise
avoid. Some drivers may simply opt to turn off new safety features. How can we influence people to
comply with new safety features?
Challenge 7: Equity – One of the main questions is the role of AVs and CVs in fostering social inclusion.
A socially inclusive society is one that provides all people and communities with the opportunity to
participate fully in political, cultural, civic, and economic life. Research suggests that inclusive societies
foster greater social cohesiveness and better standards of health, while social exclusion and lack of
community interaction are associated with poorer health outcomes and earlier death. Social exclusion
affects a wide range of individuals and communities, although certain groups, including seniors and
persons with work-limiting disabilities, are at particular risk. While many factors contribute to social
exclusion, inadequate transportation has been identified as a key contributory cause. This is because
an inability to access or use personal or public transport can leave people unable to access health
services, employment opportunities and social activities. The problem can be particularly acute in rural
areas.
AUTOMATED VEHICLES 2021R03EN
70

The issue of equity was raised for example by Mr. Litman. Although he felt that AVs were “very
unlikely” to reduce traffic congestion on city streets in the foreseeable future, he explained that
dedicated AV lanes could be implemented with the intent to reduce congestion on highways.
Nonetheless, he cautioned that publicly funded AV lanes could raise questions of social equity, as only
those who can afford to buy or use AVs would benefit from them (as noted earlier, AVs are initially
expected to be relatively expensive). This could be especially prevalent in LMICs.
The equity issues also include changes access to mobility. For example new traffic pricing policies can
lead to time segregation (less wealthy part of society will travel only in times when they can afford to
pay) and inequality in access to mobility (e.g. geofencing for less advanced vehicles) – a form of
geographical segregation.
Challenge 8: Loss of tax revenue. Nowadays vehicles are driven by people who are subject to paying
tax through the income earned partially or wholly through a driving task. When machines will drive
without professional drivers, governments will lose part of the tax revenue and will be forced to tax
“machine work”. This could lead to a scenario in which human drivers will be cheaper to hire
regardless of technological development.
Challenge 9: Privacy issues. With more intelligent and assistant systems in automated cars, more data
will be gathered from drivers and passengers. This is associated with two main risks. 1) The data are
vulnerable to abuse and could be the target of cyber-attacks. 2) The data will be source of commercial
gain for their owners but the original “providers” – road users – will have little to no gains from the
monetization of their data.
Challenge 10: Health issues. The use of AVs is expected to have impacts on the lifestyle of its users.
For example, alcohol and drugs consumption could rise significantly. As the necessity to drive is one
reason that discourages drivers from drinking excessively, it is expected that alcohol consumption will
rise by 10% when “drivers” become “passengers” of driverless vehicles.
Physical health is not the only subject of concern. Driving a car is a human activity where people
experience strong performance situations and solve complex tasks under stressful conditions.
Removing manual driving can thus lead to attempts to shift this gaining of experience into other life
domains which could become riskier. Another consequence is degeneration of human performance
skills.
Challenge 11: Resistance to change due to fear of job losses in LMICs – For example, in South Africa
more than 51% of households rely of the mini-bus taxi industry as their form of transport. The mini-
bus taxi owners, and their employed drivers, will resist the change to move toward automation. In a
country with an unemployment rate of 29% (six times the estimated average global rate), the general
workforce (including taxi drivers) is likely to resist automation at all costs. In fact, many industries,
such as the construction industry are encouraged by government to use fewer machines and to rather
use labour intensive techniques to increase job opportunities.
AUTOMATED VEHICLES 2021R03EN
71

9.4 RECOMMENDATIONS FROM THE PERSPECTIVE OF SOCIETY

For governments:
• Reinforce the Safe System approach to ensure automated vehicles are used safely.
• Apply Vision Zero thinking to automated driving.
• Avoid safety performance being used as a marketing parameter to differentiate competing
automated vehicles.
• Carefully assess the safety impacts of systems that share driving tasks between humans and
machines.
• Require reporting of safety-relevant data from automated vehicles.
• Develop and use a staged testing regime for automated vehicles.
• Establish comprehensive cybersecurity principles for automated driving.
• Ensure the functional isolation of safety-critical systems and that connectivity does not
compromise cybersecurity or safety.
• Provide clear and targeted messaging of vehicle capabilities.
For legislation:
Proactive policy, including specific rules, can provide companies the legal clarity they need to make
investment and deployment decisions and can enable governments to appropriately handle
automation technologies at each of the regulatory gates described above. Nonetheless, prematurely
codifying requirements can freeze unrealistically high or low expectations into the law in a way that
ultimately causes that law to lag rather than to lead. Duplicate, or repeat efforts to develop rules can
force developers to invest resources in lengthy legislative debates and regulatory processes.
For these reasons, informal dialogue may often be preferable to specific rules. Importantly, countries
and regions with a specific “automated driving law” are not necessarily ahead of those without one.
The US state of Michigan, for example, recently enacted a law that explicitly prohibits the operation
of automated vehicles for any purpose other than research and development testing.
As for the uniformity or flexibility of legislation there is no clear answer. Uniform regulation across
multiple jurisdictions can reduce the cost and complexity for developers of systems that will cross
national or subnational borders. Designing for one set of rules and roads is daunting enough; designing
for dozens only amplifies this challenge. Flexible regulation, however, might more easily
accommodate existing regional differences, local start-ups, and unique demonstration projects.
The choice between ex-ante regulation (particularly regulatory standards) and ex-post regulation
(particularly recalls and civil suits) implicates flexibility. Forward-looking rules provide more certainty
but less flexibility; backward-looking measures provide more flexibility but less certainty. These trade-
offs are particularly relevant to concerns raised about the liability of automated vehicle manufacturers
and associated companies for injuries related to their products. These concerns, however, likely derive
at least as much from technical uncertainty (how will these eventual products perform) as from legal
uncertainty (how will courts determine liability).
For research funding:
More democratic research is needed in AVs. The main expected target groups (people from remote
areas, women, seniors, etc.) are usually not adequately part of designing and testing of AVs which
AUTOMATED VEHICLES 2021R03EN
72

results in lower trust and acceptance in these groups and leads to some unintended consequences
(e.g. higher collision rates involving people with darker complexion as sensors are less adapted to
applicable ethnic groups).
There is also question of public money invested in research and development of AVs and infrastructure
for them. Society should express whether it is worth the money and effort or whether that money
should be invested in enhancing existing and well adapted technologies and safety features (e.g.
scenario of manual driving with ADAS).
The third main area from a social perspective is cultural research. There is wide range of issues that
can be solved by technical solutions which are mainly connected with the impacts on individuals’
abilities as a result of using AVs – loss of skills, handover, distraction, etc. These solutions can be
implemented worldwide with minimal need for adaptations (as it is mainly derived from biological
limits). But in the case of more general issues like ethics or policy making, each country will need to
do its own research to consider localised and potentially neighbouring contexts. Minimal ethical
principles must be set, but ethical principles are not the same around the world.
AUTOMATED VEHICLES 2021R03EN
73

10 CONCLUSION
Vehicle Automation – at least SAE levels 3, 4 and 5 – is still in a testing phase in most countries of the
world. Many tests at level 3 have been performed successfully on open roads without any changes to
the road infrastructure and without any impact on congestion. However, in many situations, drivers
must still take over control of the vehicle – a fall-back solution that is still allowed at level 3, but not
at level 4 or 5.
To reach the higher levels of automation, some support from the infrastructure is expected to be
needed. In particular, well-maintained lane markings and the provision of landmarks seem to be key.
But automated vehicles cannot rely solely on physical infrastructure, which has limits. For example, it
is not technically possible to predict the time of degradation of lane markings below a level where it
can impact automated operations. Therefore, digital infrastructure, including data provided through
high-definition maps and/or through vehicle-to-infrastructure connectivity, must be made available.
First insights on this digital infrastructure have been outlined and discussed in this report.
Regarding the impact on congestion, some models show that it should be slightly negative at low
penetration rates and grow positive (increase of capacity) only by 2050 or 2055.
Some studies also show that the impact on road safety should be positive at high penetration rates,
as already observed on ADAS systems, but this supposes that many human factors are considered,
such as driver overreliance or loss of driving skills. The report mentions 11 key challenges to tackle
regarding social issues to make automation a success.
AUTOMATED VEHICLES 2021R03EN
74

11 RECOMMENDATIONS
11.1 RECOMMENDATIONS FOR DECISIONS MAKERS
As in any type of business, it is important to have a user centric perspective. General digitization of
the road transport system and the introduction of highly automated vehicles needs to be undertaken
in a context that improves the everyday lives for individual and business users and with the aim of
improving safety and accessibility in a sustainable society. The focus needs to recognise that the
benefits and efficiency for individual and businesses are achieved through the collective contribution
by all actors involves. To create a balance between the possibilities and risks of digitalization,
functional safety needs to be the primary driver from the outset. Therefore, the changing transport
system needs to be tested to ensure it can fulfil the requirements of functional safety across the
challenges associated with conflicting technologies, societal conditions and human behaviours.
Wide introduction of automated vehicles, however, requires collaboration where different
competencies and actors identify challenges and opportunities together. Opportunities will include,
but are not limited to, business models, law and security, technical development, standardization, and
information management.
It has been apparent through history, and it will be true for vehicle automation, that new technology
helps to develop new solutions for improved mobility and push forward the road transport system
towards sustainable goals. Such solutions can often be more effective than the prior or more
traditional measures they will replace. Therefore, funds (both public and private) need to be
reprioritised and allocated to enable the introduction of solutions based on new technology where
sound evidence and due consideration demonstrate that the benefits are worth pursuing for the
overall benefits to communities and society more broadly.
The following are the most important recommendations for road authorities and operators:
Learn by doing
Road operators and road authorities should monitor AV testing and seek feedback from OEMs
concerning any difficulties experienced due to infrastructure – both physical and digital. A staged
testing regime for automated vehicles should be developed, involving all relevant stakeholders.
Build the digital infrastructure to complement and complete the physical infrastructure “picture”
Automated vehicles cannot rely solely on physical infrastructure, which has inherent limits. For
example, the deterioration of lane markings or impacts of weather events and natural phenomena on
infrastructure cannot always be known or remedied in a fashion that always provides certainty for
automated vehicle operations. Digital infrastructure, such as a digital environment through
connectivity and/or HD maps, will be key to allow vehicles to have sufficient clarity where sensor
information is incomplete.
Access to data and need for standardization
It is very important that authorities take responsibility for controlling and coordinating public basic
data within the road transport system, in accordance with the intentions of the basic data collection
and use.
AUTOMATED VEHICLES 2021R03EN
75

IT and communication infrastructure

Investments in IT and communication infrastructure are needed to enable connectivity throughout
the road transport system. This requires coordination of public and private investment to meet the
road transport challenges and deliver policies more effectively.
Legal aspects
Authorities need to review and present sets of new legal frameworks and commercial preconditions
for large scale introduction of automated vehicles as well as facilitate the exchange of public and
private data, both at national and international levels.
Balance between opportunities and risks
Authorities need to create conditions for continuously developing knowledge and methods for
assessing threats and balancing cyber security, especially considering the criticality and required
functionality of the road transport system.
Collaboration and collaboration areas
Frameworks that contain principles and recommendations need to be developed through
collaboration between the various public and private actors. Authorities should take the lead and
conduct relevant activities together with commercial actors and users with the objectives of
developing the next generation road transportation.
Steering mechanism
Automated vehicles will likely create new opportunities to contribute to the goals of a sustainable
road transport system. To avoid negative system effects and to obtain the desired changes, policy
instruments may need to be developed for funding both the transition and planned ultimate scenarios
(e.g. taxes, fees, and other regulations).
Financing of digital transformation for the road transport system
Further development of cost-benefit analysis needs to be undertaken in order to estimate or better
describe the possible and likely societal benefits and associated costs in order to asses and prioritize
various investments on various components, in particular communications infrastructure, road
pavement marking, digital twins, HD Maps etc.
Take account of social issues

• Reinforce the Safe System approach to ensure automated vehicles are used safely.
• Apply Vision Zero thinking to automated driving.
• Avoid safety performance being used to market competing automated vehicles.
• Carefully assess the safety impacts of systems that share driving tasks between humans and
machines.
• Require reporting of safety-relevant data from automated vehicles.
• Establish comprehensive cybersecurity principles for automated driving.
• Ensure the functional isolation of safety-critical systems and that connectivity does not
compromise cybersecurity or safety.
• Provide clear and targeted messaging of vehicle capabilities.
AUTOMATED VEHICLES 2021R03EN
76

11.2 RECOMMENDATIONS FOR ROAD OPERATORS AND ROAD AUTHORITIES

11.2.1 Short term
• Follow tests and obtain feedback from OEMs about any difficulties associated with the
infrastructure
• Require reporting of safety-relevant data from automated vehicles
• Ensure a proper level of maintenance of lane markings and critical physical road infrastructure
to support vehicle positioning and guidance.
• Test and develop infrastructure support through I2V connectivity in some critical zones (e.g.
road works)
• Ensure the functional isolation of safety-critical systems and that connectivity does not
compromise cybersecurity or safety.
11.2.2 Mid-and long term

• Adapt road design rules and maintenance strategies that respond to lessons learnt
• The physical infrastructure will become increasingly digitised. A significant challenge is to build
this digital infrastructure and to keep it up to date (in particular HD maps)
• Support harmonization of traffic signs and markings
• Automated shuttles need to be considered in an urban and rural environment

11.3 RECOMMENDATIONS FOR LMICS

Many of the changes being brought on by the technological advancements in mobility, will prove to
be very disruptive and potentially have a negative social and socio-economic impact in LMICs in
particular. As discussed in Chapter 9 (Social Issues and the Consideration of Human Factors), many
LMICs have extremely high unemployment rates (29% - In the case of South Africa, which is six times
the global average), and the retention and creation of jobs is a priority for the their governments.
Therefore, any form of automation will be a threat to humans, in terms of job losses and resulting
livelihood. However, LMICs cannot afford to get left behind the global community. It is the
responsibility not only of governments but the transport industry at large to prepare for these changes
and to not be caught off-guard.
There are many automotive industry producers in LMICs (Africa, Asia and South America) and they
export most of the vehicles they produce to some of the more developed countries (e.g. 60% in the
case of South Africa). Due to the technological advancements in the global automotive industry,
export markets across the world are all at various stages of testing and deploying CAVs. These LMIC
automotive producers will be expected to meet the demands of the export market in the production
of these vehicles. It is therefore imperative that LMICs prepare well in advance to be ready to meet
the increasing demand for new technologies and automation.
Irresponsible behaviour and human factors account for up to 90% of serious accidents in South Africa,
it is therefore essential that the benefits of technology in the form of CAV are explored to remove the
human element as far as practicable. There are many socio-economic challenges that need to be
addressed first, but the benefits of CAV technology cannot be ignored. If the LMICs are to strive for
“vision zero,” in line with many developed countries, the benefits of CAV technology need to be fast-
tracked by governments, road authorities and the transport industry.
AUTOMATED VEHICLES 2021R03EN
77

It is therefore recommended that LMICs:

• Start small, with pilot deployments on the most mature services as piloted in the developed
countries
• Start with safety services first as road safety is the biggest motivation for deployment
• Explore the law enforcement possibilities of CAV and C-ITS to assist in improved road user
behaviour.

11.4 RECOMMENDATIONS FOR PIARC

Automation will continue to be a hot topic in the next decade for the road transport and mobility
community. It is recommended that work is continued, with a new Task Force, until next World Road
Congress in 2023.
A few topics to be considered in greater depth are, among others:
• Cost/Benefit analysis and financing models
• Necessities and Luxuries - “must haves” and “nice to haves” from the perspective of society
• Universal aspects should be discouraged or prohibited from being used for competitive
marketing of vehicles and features
• Minimal ethical principles and requirements
• Minimal data protection and cybersecurity principles and requirements
• Principles for policymaking and helping end-users
After the world congress of 2023, the topic could be taken over by a dedicated Technical Committee.
AUTOMATED VEHICLES 2021R03EN
78

12 GLOSSARY

ACC Adaptive Cruise Control

ADAS Advanced Driver Assistance Systems

ADS Automated Driving System

API Application Program Interface

AV Automated Vehicle

CAV Connect and Automated Vehicle

C-ITS Cooperative Intelligent Transport System

CV Connected Vehicle

DCAT-AP Data Catalogue Vocabulary – Application Profile

DGNSS Differential GNSS

E2E End-To-End

Galileo European Global Satellite Navigation System

GIS Geographical Information System

GNSS Global Navigation Satellite System

HMI Human Machine Interface

I2V Infrastructure to Vehicle (Communications)

ISA Intelligent Speed Assistance

ISAD Infrastructure Support levels for Automated Driving

LMICs Lower and Middle Income Countries

LTA Left Turn Assist

METR Management of Electronic Traffic Regulations

MUTCD Manual on Uniform Traffic Control Devices

NAP National Access Points

AUTOMATED VEHICLES 2021R03EN
79

ODD Operational Design Domain

OEM Original Equipment Manufacturers

PPP Precise Point Positioning

Q-WARN Queue Warning

RLVW Red Light Violation Warning

RNO Road Network Operations

RTK Real-Time Kinematic

SAE Society of Automotive Engineers

SBAS Space Based Augmentation Systems

SF-PPP Single Frequency Precise Point Positioning

SLAM Simultaneous Localization and Mapping

SPaT Signal Phase and Timing

SWIW Spot Weather Impact Warning

TMC / TCC Traffic Management Center / Traffic Control Center

TSR Traffic Sign Recognition

V2I Vehicle to Infrastructure (Communications)

V2V Vehicles to Vehicle (Communications)

V2X Vehicle to Other (Communications)

WZDx Work Zone Data Exchange

AUTOMATED VEHICLES 2021R03EN
80

13 REFERENCES

[1] PIARC TF. C.1 on Infrastructure Security, “Security of Road Infrastructure – Final Report,” PAIRC,
Paris, 2019.

[2] “HTTPS://WWW.UNECE.ORG/TRANS/ROADSAFE/EG_ROAD_SIGNS_SIGNALS.HTML”.

[3] Austroads, “AP-R580-15: Implications of Traffic Sign Recognition Systems for Road Operations”.

[4] City of Calgary, “Calgary Automous Shuttle - ACATS final report,” Calgary, 2019.

[5] “https://ops.fhwa.dot.gov/weather/q1_roadimpact.htm?fbclid=IwAR2uF2hCzguNvFuH3kqXis-
irfqZmAnffVD5MABwChx119kP3JKinYGe23I”.

[6] “https://www.engadget.com/2018/05/08/waymo-snow-navigation/”.

[7] “http://wavelab.uwaterloo.ca/?weblizar_portfolio=real-time-filtering-of-snow-from-lidar-
point-clouds”.

[8] National Researsh Council Canada, funded by Transport Canada's ecoTECHNOLOGY for Vehicles
Program, “Review of cooperative truck platonning systems,” SAE International, Ottawa, Canada,
2014.

[9] J. W. Matthew Yarnold, “Truck Platoon Impacts on Steel Girder Bridges,” Journal of Bridge
Engineering, vol. 24, no. 7, 2019.

[10] Peleton, “Overview on Driver-Assitive Truck Platooning,” in Highway Engineering Conference,

Illinois, 2019.

[11] J. National Institue for Infastrucutre and Land Management.

[12] S. o. C. Standing Senate Committee on Transport and Communications, “Driving Change:

Technology and the future of the automated vehicle,” Ottawa, 2018.

[13] N. A. o. Sciences, “Reviewing the National Commitment to the Interstate Highway System,” The
National Academies Press, Washington, DC, 2018.

[14] C. T. Systems, “Future Proofing Infrastructure for Connected and Automated Vehicles,” 2017.

[15] Austroads, “AP-R580-15: Implications of Traffic Sign Recognition Systems for Road Operators”.

[16] “https://blog.csdn.net/qq_17256689/article/details/100037699”.

[17] “Leading the world to 5G:Cellular Vehicle-to-Everything (C-V2X) technologies”.

AUTOMATED VEHICLES 2021R03EN
81

[18] “https://www.transportation.gov/av/data/wzdx”.

[19] “Centre for Connected & Autonomous Vehicles,” 2018.

[20] SIP (the Cross-Ministerial Strategic Innovation Promotion Program), “SIP-adus: Project Reports,
2014-2018 - Automated Driving for Universal Services,” 2019.

[21] PIARC - TF B.1, “Connected Vehicles – Challenges and Opportunities for Road Operators,” 2019.

[22] Huawei, “Ten application scenarios in the 5G era White Paper”.

[23] Austraods, “Infrastructure changes to support automated vehicles on rural and metropolitan
highways and freeways: Audit Specification (Module 2)”.

[24] Austroads, “CAV Open Data Recommendations”.

[25] S. e. al, “Implementation and Evaluation of Local Dynamic Map in Safety Driving Systems,”
Journal of Transportation Technologies, 2015.

[26] Austroads, “Infrastructure changes to support automated vehicles on rural and metropolitan
highways and freeways: Audit Specification (Module 1)”.

[27] “https://inspire.ec.europa.eu/documents/inspire-efficient-way-share-european-spatial-data”.

[28] Austroads, “Assessment of Key Road Operator Actions to Support Automated Vehicles”.

[29] “https://inspire.ec.europa.eu/sites/default/files/presentations/0930_tn-
its_inspire_20.09.2018_v0.1.pdf”.

[30] “https://www.cdbb.cam.ac.uk/system/files/documents/TheGeminiPrinciples.pdf”.

[31] E. S. A. T. a. v. A. B. Farah H, “Infrastructure for Automated and Connected Driving: State of the
Art and Future Research Direction”.

[32] CARTRE, “Position Paper on Physical and Digital Infrastructure (PDI),” Brussels, October 2018.

[33] e. a. Anna Carreras, “Road infrastructure support levels for automated driving,” 25th ITS World
Congress, Copenhagen, Denmark, September 2018.

[34] European Commission, “DCAT Application profile for data portals in Europe (DCAT-AP),”
Brussels, Belgium, 2015.

[35] Cooperation AT, DE and NL, “SPA – Coordinated Metadata Catalogue,” 2015.
AUTOMATED VEHICLES 2021R03EN
82

[36] ERTRAC Working Group “Connectivity and Automated Driving”, “Connected Automated Driving
Roadmap,” Brussels, Belgium, 2019.

[37] A. Somers, “Connected and Automated Vehicles (CAV) Open Data Recommendations,”
Austroads Ltd., Sydney, Australia, 2018.

[38] Official Journal of the European Union, “Commission delegated regulation (EU) No 886/2013 of
15 May 2013 supplementing Directive 2010/40/EU of the European Parliament and of the
Council with regard to data and procedures for the provision,” Brussels, Belgium, 2013.

[39] Bundesminister für Verkehr und digitale Infrastruktur, “Ethik-Kommission – Automatisiertes und
vernetztes Fahren,” Berlin, Germany, 2017.

[40] European Commission, “C-ITS Platform – Final Report,” Brussels, Belgium, 2016.

[41] Data Task Force: Memorandum of Understanding (“MoU”), “Increasing road safety by sharing
road safety related data in public and private cooperation,” Eindhoven, Netherlands, 2019.

[42] M. &. M. N. &. K. S. &. H. S. &. V. P. &. B. F. Hartmann, “Impact of Automated Vehicles on Capacity
of the German Freeway Network.,” 2017.

[43] UN, “World Urbanization Prospects,” 2018.

[44] Ricardo Energy&Environment, “Cost Analysis of V2I Deployment,” 5G Automotive Association,

August 2020.

[45] N. &. al., “Challenges to Human Drivers in Increasingly Automated Vehicles,” 2018.

[46] W. e. al., “Attentional Models of Multitask Pilot Performance Using,” 2003.

[47] Fitts, “Reflections on the 1951 Fitts List: Do Humans Believe Now that Machines Surpass them?,”
1951.

[48] G. Cummings, “The New Statistics: Why and How,” 2014.

[49] D. D. Joost de Winter, “A surge of p-values between 0.041 and 0.049 in recent decades,” 2014.

[50] M. S. BRANDON SCHOETTLE, “A PRELIMINARY ANALYSIS OF REAL-WORLD CRASHES INVOLVING

SELF-DRIVING VEHICLES,” 2015.

[51] W. G. &. E. Mazzae, “Driver Distraction Research: Past Present and Future,” NHTSA , 2000.

[52] R. Parasuraman, T. Sheridan and C. Wickens, “A model for types and levels of human interaction
with automation. IEEE Trans. Syst. Man Cybern. Part A Syst. Hum. 30(3), 286-297,” IEEE, 2000.
AUTOMATED VEHICLES 2021R03EN
83

[53] Comte, 1998.

[54] M. Hjälmdahl and A. Varhelyi, “Validation of in-car observations, a method for driver
assessment,” 2004.

[55] H. &. Kopf, “Modelling Driver Behaviour in Automotive Environments,” 2001.

[56] R. -. B. a. Parker, “Behavioural adaptations to Acc,” 2004.

[57] N. Y. M. a. M. B. Stanton, “Drive-by-wire: the case of mental workload and the ability of the
driver to reclaim control,” 1997.

[58] L. e. al, 2008.

[59] R. J. Loon and M. H.Martens, “Automated Driving and its Effect on the Safety Ecosystem: How
do Compatibility Issues Affect the Transition Period?,” 2015.

[60] G. e. al., “Driving next to automated vehicle platoons: How do short time headways influence
non-platoon drivers’ longitudinal control?,” 2014.

[61] T. &. al., “CityMobil: Human Factor Issues Regarding Highly-automated Vehicles on an eLane,”
2009.

[62] D. I. Bianbridge, “Computers and Copyright,” 1987.

[63] R. P. P.A.Hancock, “Human factors and safety in the design of intelligent vehicle-highway
systems (IVHS),” Journal of Safety Research, 1992.

[64] R. Parasuraman, “Human-Computer Monitoring,” 1987.

[65] C. M. &. P. H. A. Rudin-Brown, “Behavioural adaptation to adaptive cruise control (ACC):

Implications for preventive strategies,” 2004.

[66] M. S. Y. &. N. Stanton, “What's skill got to do with it? Vehicle Automation and driver mental
workload,” 2007.

[67] K. &. Moyer, “Integration of driver in-vehicle ITS information,” FHA, 2000.

[68] Bloomberg Philanthropies, “Taming the Autonomous Vehicle: A Primer for Cities,” 2017.

[69] SAE J3016, “Surface Vehicle Recommended Practice — Taxonomy and Definitions for Terms
Related to Driving Automation Systems for On-Road Motor Vehicles,” July 2018.
Copyright by the World Road Association. All rights reserved.

World Road Association (PIARC)

La Grande Arche, Paroi Sud, 5e étage, F-92055 La Défense cedex