12th IFAC Symposium on Robot Control

12th IFAC Symposium

Budapest, on Robot
Hungary, August Control
27-30, 2018
12th IFAC Symposium
Budapest, on Robot
Hungary, August Control
27-30, Available online at www.sciencedirect.com
2018
12th IFAC Symposium
Budapest, on Robot
Hungary, August Control
27-30, 2018
Budapest, Hungary, August 27-30, 2018
ScienceDirect
IFAC PapersOnLine 51-22 (2018) 132–137
Modeling, Simulation and Control of a Pipe Inspection Mobile Robot
Modeling, Simulation
with and Control
an Active of a Pipe Inspection
Adaptation System Mobile Robot
Modeling, Simulation and Control of a Pipe Inspection Mobile Robot
with and
Modeling, Simulation an Active
ControlAdaptation
of a Pipe System
Inspection Mobile Robot
with an Active Adaptation System
withMichał
an Active Adaptation
Ciszewski*, System
Tomasz Buratowski**
Michał Ciszewski*, Tomasz Buratowski**
Michał Ciszewski*, Tomasz Buratowski**
Mariusz Giergiel***
Michał Ciszewski*,Mariusz Tomasz Buratowski**
Giergiel***
Michał Ciszewski*, Tomasz Buratowski**
Mariusz Giergiel***
*AGH University of Science and Technology, Mariusz Faculty of Mechanical Engineering and Robotics,
Giergiel***
*AGH University of Science30,
Al. Mickiewicza and Technology,
30-059 Kraków, Faculty
Poland, of (e-mail:
Mechanical Engineering and Robotics,
[email protected]).
*AGH UniversityAl. of Science30,
Mickiewicza and Technology,
30-059 Kraków, Faculty
Poland, of (e-mail:
Mechanical Engineering and Robotics,
[email protected]).
** AGHUniversity
*AGH University
Al. ofofScience
MickiewiczaScience and
and
30, Technology,
Technology,
30-059 Kraków, Faculty
Faculty
Poland, ofof(e-mail:
Mechanical
Mechanical Engineeringand
Engineering
[email protected]). andRobotics,
Robotics,
** AGH University
Al. of Science
Mickiewicza 30, and Technology,
30-059 Kraków, Faculty (e-mail:
Poland, of Mechanical Engineering and Robotics,
[email protected]).
Al.
** AGH University Mickiewicza
of Science 30, 30-059 Kraków,
and Technology, Poland, (e-mail:
Faculty (e-mail:
of Mechanical [email protected]).
Engineering and Robotics,
*** Al. Mickiewicza
AGHUniversity
University ofScience30, 30-059
Science Kraków, Poland,
andTechnology,
Technology, Faculty(e-mail: [email protected]).
ofMechanical
Mechanical Engineeringand andRobotics,
Robotics,
** AGH Al. of
Mickiewicza 30, and
30-059 Kraków, Faculty
Poland, of Engineering
[email protected]).
***Mickiewicza
Al. AGH University 30, of Science
30-059 Kraków,and Technology,
Poland, (Tel:Faculty
+48 12 of617
Mechanical
36 75; Engineering
e-mail: and Robotics,
[email protected]).
***Mickiewicza Al. Mickiewicza
AGH University 30,
of Science 30-059 Kraków,
and Technology, Poland, (e-mail: [email protected]).
Al. 30, 30-059 Kraków, Poland, (Tel:Faculty
+48 12of617 Mechanical Engineering
36 75; e-mail: and Robotics,
[email protected]).
***Mickiewicza
Al. AGH University of Science
30, 30-059 Kraków,and Technology,
Poland, (Tel:Faculty
+48 12of617 Mechanical Engineering
36 75; e-mail: and Robotics,
[email protected]).
Al. Mickiewicza 30, 30-059 Kraków, Poland, (Tel: +48 12 617 36 75; e-mail: [email protected]).
Abstract: In this paper, a tracked inspection mobile robot with an active adaptation system is presented. It
Abstract:
can be usedIn for
thisvisual
paper,inspection
a tracked of inspection mobile robot
various pipelines. with an active
Mathematical adaptation
modeling system is mechanisms
of pedipulator presented. It
Abstract:
can be In for
used thisvisual
paper,inspection
a tracked of inspection
various mobile robot
pipelines. with an active
Mathematical adaptation
modeling of system is mechanisms
pedipulator presented. It
that
can allow
Abstract:
be adaptation
In for
used thisvisual of the
paper,inspection robot
a tracked of to different
inspection
various shapes and
mobile robot
pipelines. sizes of pipes
with an active
Mathematical is described
adaptation
modeling with focus on forward
system is mechanisms
of pedipulator presented. It
that
and allow
inverse adaptation
kinematics of the robot
methods, to different
implemented shapes
in and
MATLAB sizes of pipes
software. is described
Application with
of a focus
custom ontrajectory
forward
can be
that usedadaptation
allow for visualofinspection
the robot of
to various
different pipelines.
shapes Mathematical
and sizes of pipesmodeling
is of pedipulator
described with focus mechanisms
on forward
and
that inverse
planning kinematics
allowalgorithm
adaptation methods,
forofthe robot implemented
thepedipulators is shown
to different in with
shapesMATLAB sizessoftware.
numerical
and of and Application
pipesgraphical
is described of
validation.
witha custom ontrajectory
Co-simulations
focus forward
and
planninginverse kinematics
algorithm for methods,
the implemented
pedipulators is shownin with
MATLAB
numerical software.
and Application
graphical of
validation. a custom trajectory
Co-simulations
of
andthe
planning robot
inverse motion,for
kinematics
algorithm prepared
methods,
the in V-REP
implemented
pedipulators isand MATLAB
shownin with
MATLAB software
numerical are graphical
software.
and performed
Application forofhorizontal
validation. a custom and rough
trajectory
Co-simulations
of the robot
surfaces
planning and motion,
horizontal
algorithm prepared
for thepipe in V-REPis
segments
pedipulators and
with MATLAB
connections
shown with software
and
numerical are graphical
reducers.
and performed
A prototypefor horizontal
validation.of the and with
robot
Co-simulations rougha
of the robot
surfaces and motion,
horizontal prepared
pipe in V-REPwith
segments and connections
MATLAB software and are performed
reducers. A prototypefor horizontal
of the and with
robot rougha
model-based
of the robot
surfaces and control
motion,
horizontal system
prepared
pipe issegments
intested
V-REP and laboratory
and
with MATLAB experiments
connections software
and are performed
are
reducers. compared
A prototypeto
forthe simulation
horizontal
of the andresults.
robot rougha
with
model-based
Different
surfaces and control
motion
horizontal system
scenarios of
pipe issegments
thetested
robot andare
withlaboratory
shown withexperiments
connections focus
and on are compared
adaptation
reducers. A to theofsimulation
capabilities
prototype of the
the motion
robot results.
unita
with
model-based
Different control
motion systemofisthe
scenarios tested
robotand arelaboratory
shown with experiments are compared
focus on adaptation to the simulation
capabilities of the motion results.
unit
to work
model-based
Different environment.
control system is tested and laboratory experiments are compared
motion scenarios of the robot are shown with focus on adaptation capabilities of the motion unit to the simulation results.
to work environment.
Different motion scenarios of the robotofare shown with focusHosting
on adaptation capabilities
© work
to 2018, environment.
Keywords: IFAC (International
Mobile Federation
robots, Pipelines, Robot Automatic Control)
kinematics, Computer by Elsevier
simulation, Ltd. All of
Numerical
the motion
rights reserved.
simulation,
unit
CAD
to work environment.
Keywords: Mobile robots, Pipelines, Robot kinematics, Computer simulation, Numerical simulation, CAD
models.
Keywords: Mobile robots, Pipelines, Robot kinematics, Computer simulation, Numerical simulation, CAD
models.
Keywords: Mobile robots, Pipelines, Robot kinematics, Computer simulation, Numerical simulation, CAD
models.
models. in horizontal pipe segments of rectangular or circular cross-
1. INTRODUCTION in horizontal pipe segments
1. INTRODUCTION section, beyond
in horizontal Ø300
pipe mm of of
segments of
rectangular
diameter
rectangular(Inuktunor circular
2017). cross-
or circular cross-
1. INTRODUCTION
Inspection of pipelines has several objectives. After building a in section, beyond
horizontal Ø300
pipe mm
segments of diameter
of rectangular(Inuktun 2017).
or circular cross-
1. INTRODUCTION section,
Several beyond
robots Ø300
that mm
operate of
in diameter
vertical (Inuktun
pipes are 2017).
also present in
Inspection
new pipeline,of pipelines
it is has by
required several objectives.
regulations that After
the pipebuilding
must bea Several
section, beyond
robots Ø300
that mm
operate of
in diameter
vertical (Inuktun
pipes are 2017).
also present in
Inspection
new pipeline,of pipelines
it device, has
is required several
by objectives.
regulations that theAfterpipebuilding a the
must bea Several market. The Inuktun Versatrax Vertical tracked robot
inspected
Inspection by
of a
pipelines equipped
has several with a
objectives.camera. After In addition,
building the robotsThe
market. thatInuktun
operate in vertical pipes
Versatrax Verticalare also
trackedpresent in
robot
new pipeline,
inspected by ait device,
is required by regulations
equipped with a that the pipe
camera. In must be features
addition, Several pantograph
robots that adjustment
operate in mechanism
vertical pipes and
are can
also be used
present in
in
inclination
new pipeline, of aitthe
is pipe has
required to be
by precisely
regulations measured.
that the pipe With
mustthe the
be features market. The
pantograph Inuktun
adjustment Versatrax
mechanism Vertical tracked
and can robot
inspected
inclination byof the device,
pipe equipped
has to be with a camera.
precisely measured. In addition,
With the vertical
the market.
features pipes with Inuktun
The
pantograph diameters
adjustment in range
Versatrax
mechanism Ø200÷300
Verticaland mmbe
tracked
can be
used
(Inuktunin
robot
used in
initial
inspected inspection,
inclination byof athe engineering
device,
pipe equipped
has errors can
with a camera.
to be precisely be
measured. found and
the vertical
In addition,
With and pipes with diameters in range Ø200÷300 mmbe (Inuktun
initial inspection, engineering errors can be found 2017).
featuresAnother
pantograph solution is provided
adjustment mechanism by Neovision
and can company
used in
eliminated
inclination before
of the exploitation
pipe has to be of an industrial
precisely facility.
measured. Other
With the vertical
2017). pipes
Another with diameters
solution is in range
provided Ø200÷300
by Neovision mmcompany
(Inuktun
initial
eliminatedinspection, engineeringof errors
before techniques
exploitation can be
an periodical
industrial foundOther
facility. and vertical
that offers a tracked
pipes with robot, Jetty,
diameters in rangeintended
Ø200÷300 for inspection
mmcompany and
(Inuktun
uses of inspection
initial inspection, include
engineeringof errors can be checks of pipe
foundOtherand that 2017). Another
offers solution is provided by Neovision
eliminated
uses of before techniques
inspection exploitation include an periodical
industrial facility.
checks of pipe cleaning
2017). ofavertical
Another tracked robot,
ducts
solution is Jetty,
(NEOVISION
provided intended
by forn.d.).
s.r.o. inspection
Neovision Itcompanyand
features
condition
eliminated orbefore
documentation
exploitation ofinclude
old pipes.
of an industrial facility. that offers avertical
trackedductsrobot, Jetty, intended forn.d.).
inspection and
uses of inspection
condition or techniques
documentation of old periodical
pipes. checks ofOtherpipe cleaning
six
that track
offers of
drive modules,
avertical
tracked robot, (NEOVISION
mounted
Jetty, intended s.r.o.
on pantograph It features
mechanisms
forn.d.).
inspection and
uses of inspection techniquesofinclude periodical checks of pipe six cleaning
track of
drive ducts
modules, (NEOVISION
mounted on s.r.o.
pantograph It features
mechanisms
condition
Due to theorfact documentation
that pipelines old not
are pipes.accessible by operators and provides
cleaning of inspection
vertical ducts of circular
(NEOVISION and rectangular
s.r.o. n.d.). ducts
It with
features
condition six track
and providesdriveinspection
modules, of mounted
circular on
and pantograph
rectangular mechanisms
ducts with
Due
for to theor
direct
documentation
fact
visual that pipelines
assessment,
of are
old not
pipes.
robotic accessible
inspection by systems are diameters
operators six track
and provides from
drive Ø400 mm
modules,
inspection to Ø1300
mounted
of circular on
and mm in extended
pantograph
rectangular mechanisms
ducts witharm
Due
for to the
direct fact
visual that pipelines
assessment, are not
robotic accessible
inspection by operators
systems are diameters
version. from Ø400 mm to Ø1300 mm in extended arm
applied
Due
for to the
direct forvisual
search
fact that of leakages,
pipelines
assessment, are not
roboticdiscontinuities,
accessible
inspection by corrosion,
operators
systems are and provides
diameters frominspection
Ø400 mm of circular
to Ø1300 and rectangular
mm in extended ducts witharm
applied
blockages forandsearch
other of leakages,
defects. The discontinuities,
robots have to be corrosion,
designed to version.
diameters from Ø400 mm to Ø1300 mm in extended arm
for direct visual
applied forandsearch assessment,
of leakages, robotic inspection
discontinuities, systems are
corrosion, version.
blockages
move
applied in for other defects.
changeable
search of The robots
environment
leakages, have to
because
discontinuities, be designed
wide variety
corrosion,to Application
of version.
of control systems for mobile robots that allow
blockages and other defects. The robots have to be designed Application
to track adaptationof control
for pipes systems for
with variable mobile robots
diameters that
is that allow
a complex
move
pipelinesin changeable environment because wide variety of Application of control systems for mobile robotsis allow
move in exist
blockages
pipelines
and other
exist
in industry.
changeable
in
defects.
industry.
Horizontal
The robots
environment
Horizontal
pipelines
have to
because
pipelines widearedesigned
be
are
relatively
variety to track
of
relatively task. adaptation
An
Application analytical
of for approach
pipes
control withfor
systems variable diameters
parallelogram
for mobile robot
robots a complex
thatthat can
allow
easy to inspect,
move in exist changeable but inclined
environment or vertical
because pipe segments
wide with track adaptation for pipes with variable diameters is a complex
pipelines
easy to inspect, inbut
industry.
inclined Horizontal
or verticalpipelines
pipe arevariety
segments
of task.
relatively
with adjustAn
track analytical
traction
adaptation force
for approach
during
pipes with for
motionparallelogram
variable indiameters
pipes was robot
is that can
shown
a complex by
junctions
pipelines can be
exist challenging
inbut
industry. forvertical
Horizontal designers
pipelines ofsegments
robot motion
are relatively task. An analytical
forceapproach for parallelogram robot that can
easy to inspect,
junctions can be inclined
challenging or
for designers pipeof robot with adjust
motion (Zhang
task. Antraction
& Yan 2007).
analytical during
In this
approach motion
research
for inrobot
a pipes
parallelogram forwas shown
circular
robot by
cross-
that can
units
easy toandinspect,
control system engineers. adjust traction force In during motion ainrobot pipesforwas shown by
junctions
units can bebut
and control
inclined
challenging
system
or
forvertical
engineers. designers pipeofsegments
robot motion with (Zhang
section
adjust & Yanis2007).
pipes
traction shown,
force thismotion
but
during research
motioninainother pipestypescircular
was ofshown cross-
pipelines
by
junctions can be challenging for designers of robot motion (Zhang & Yan 2007). In this research robot for circular cross-
units and control
At present, there system
exist many constructions of pipe inspection section
engineers. is
(Zhang
section &pipes
Yanis
not investigated.
pipes is
shown,
2007).
shown, In but
thismotion
but researchin
motion inaother
robot types
other
of pipelines
for circular
types of cross-
pipelines
units
At and control
present, system engineers. is not investigated.
robots. The there exist many
most popular onesconstructions
are wheeled platformsof pipe inspection
intended is section pipes is shown, but motion in other types of pipelines
At present,
robots. there
Themonitoring exist
most popular manyones constructions
are wheeled of pipe
platforms inspection
intended In not
thisinvestigated.
paper, a tracked inspection robot with an active
for
At visual
present, there exist of
manyhorizontal pipelines,
constructions of equipped
pipe with
inspection is
In not
thisinvestigated.
paper, a tracked inspectionwith robot
robots.
for The
visual most
monitoring popular ones
of such are
horizontal wheeled
pipelines, platforms
equipped intended adaptation
with In this paper, mechanism
a tracked is presented
inspection focuswith
robot
an active
on modeling,
with an active
Pan-Tilt-Zoom
robots. The most
for visual monitoring cameras
popular ones
of such as
are
horizontal ABE
wheeledGroup
pipelines, products
platforms (ABE
intended
equipped(ABE with In adaptation mechanism is presented with focus on modeling,
Pan-Tilt-Zoom cameras as ABE Group products simulations
this paper,
adaptation and acontrol
mechanism trackedis system
presented thatwith
inspection can servewith
robot
focus as
on amodeling,
anversatile
active
Group
for 2017).
Pan-Tilt-Zoom Tracked
visual monitoring cameras motion
of such units
horizontal
as ABEare also
pipelines,
Group used in the
equipped(ABE
products robots.
with simulations and control system thattowith
can serve as amodeling,
versatile
Group 2017).adjustable
Tracked motion inspection
adaptation device.
mechanism The robot
is is able
presented move in
focus pipes
on and ducts
A manually
Pan-Tilt-Zoom
Group cameras
2017).adjustable
Tracked such units
chassis
motion forABE
as
units
are also
motion Group
are also inused
used
in thetypes
different
products robots.
(ABE
in thetypes
robots.
simulationsdevice.
of inspection
with round
and control
and The robot
rectangular
systemis ablethatto can
cross-section move serve
thatinare as aand
pipes versatile
horizontal ducts
or
A manually chassis for motion in different of simulations
inspection and
device. control system
The robotcross-section that
is able to movecan serve as a versatile
pipelines
Group
A manually is present
2017). Tracked
adjustable in Inuktun
motion Versatrax
chassis units are also
for motion 300, which
inused in the
different can move
robots.
types of with round
vertical,
inspection withandbends
device.rectangular
androbot
The interconnections.
is able to move thatin
in
pipes
are and ducts
horizontal
pipes and or
ducts
pipelines
A manually is present
adjustable in Inuktun Versatrax
chassis Versatrax
for motion300, 300, which
in different can move
types with round
of vertical, and rectangular cross-section that are horizontal or
pipelines is present in Inuktun which can move with round withandbends and interconnections.
rectangular cross-section that are horizontal or
pipelines is present in Inuktun Versatrax 300, which can move vertical, with bends and interconnections.
vertical, with bends and interconnections.
Copyright
2405-8963 ©© 2018,
2018 IFAC 132Hosting by Elsevier Ltd. All rights reserved.
IFAC (International Federation of Automatic Control)
Copyright
Peer review©under
2018 responsibility
IFAC 132Control.
of International Federation of Automatic
Copyright © 2018 IFAC
10.1016/j.ifacol.2018.11.530 132
Copyright © 2018 IFAC 132
IFAC SYROCO 2018
Budapest, Hungary, August 27-30, 2018 Michał Ciszewski et al. / IFAC PapersOnLine 51-22 (2018) 132–137 133

2. MECHANICAL STRUCTURE OF THE ROBOT tasks (Wang et al. 2010). Selection of a proper modeling
method is dependent on the analyzed structure, since some
The design of the robot adaptable driving mechanism is based efficient methods for open kinematic chains may lose
on two pedipulators that control the pose of track drive convergence near singular positions (Dulęba & Opałka 2013).
modules. Virtual mechanical model of the robot is shown in
Fig. 1. Each pedipulator is a subsystem, attached to the body Mathematical modeling of the pipe inspection robot was
(2) that consists of two actuated rings (3, 4), front and rear arm divided into three main sections: kinematic motion model for
(5, 6) and a track drive module (1). The rings, with an axis of even surfaces and parallelly oriented tracks, described in
rotation in the center of the robot body are driven by (Ciszewski et al. 2013), dynamic model of the robot motion in
servomotors, from which torque is transferred by sprockets of water that was shown in (Ciszewski et al. 2014) and kinematic
internal meshing gear transmissions. Front and rear arms are modeling of pedipulator motion necessary for adaptation of the
mounted to the rings by revolute joints. Other sides of the arms robot chassis to changeable work environment. The last
are attached to the track drive module by rotary joints. The rear modeling task, presented in this paper, involves a novel,
arm is equipped with an additional servomotor that sets unconventional modeling approach focused on control of
orientation of the track module with respect to the arm. robot servomotors to attain desired poses of two track drive
modules. Inverse kinematics solution is implemented
iteratively in an original algorithm for calculation of
pedipulators transformation trajectories.
The pedipulator mechanism can be treated as planar, due to the
fact that the rotational joints axes are parallel. One pedipulator
consists of six rotary joints of which three are actuated (see
Fig. 2). The mechanism was divided into two planar
manipulators with two and three degrees of freedom that
represent front and rear parts of the pedipulator to facilitate
modeling tasks.

Fig. 1. Robot model: 1 – track drive module; 2 – robot body;

3 – front rotating ring; 4 – rear rotating ring; 5 – front arm;
6 – rear arm; 7 – camera; 8 – light; 9 – waterproof connector.

By assembling two pedipulator modules, motion unit of the

robot can adapt to different types of pipes and thus realize
dexterous manipulations, resembling capabilities of human
legs. The robot is equipped with eight drives in total, from
which six servomotors are responsible for setting position and
orientation of two independently motorized track drive
modules.
The robot model was created in Autodesk Inventor software.
Fig. 2. Kinematic model of pedipulator: 𝜃𝜃𝑟𝑟1 – rear ring (5)
In the CAD 3D model creation process, optimization of
rotation angle; 𝜃𝜃𝑓𝑓1 – front ring (2) rotation angle; 𝜃𝜃𝑟𝑟3 – rear
construction, drive selection and initial validation of
operational capabilities were conducted with the assumptions arm rotation angle with respect to track drive module (1);
of maximum versatility, compactness and watertightness. 𝜃𝜃𝑟𝑟2 , 𝜃𝜃𝑓𝑓2 – unactuated joints rotation angles.

3. MATHEMATICAL MODELING OF THE ROBOT Forward kinematic equations for the pedipulators were
formulated for 2-DOF (front) and 3-DOF (rear) manipulators
Mathematical modeling of redundant robots and closed extracted from the structure with usage of Denavit-Hartenberg
kinematic chains usually requires special techniques to notation. Additionally, it was assumed that the manipulators
provide, efficient and valid solution for inverse kinematics have to move their end-effectors concurrently to fulfil
task. Models for redundant robots with open kinematic chain constraints of the closed kinematic chain. It is realized by
can be formulated with usage of Jacobian transpose, least- equating x and y positions of both D-H transformation matrices
squares methods, but to avoid singular poses and improve (Ciszewski et al. 2016).
robustness, custom algorithms are adapted for specific control

133
IFAC SYROCO 2018
Budapest, Hungary, August 27-30, 2018 Michał Ciszewski et al. / IFAC PapersOnLine 51-22 (2018) 132–137
134

Inverse kinematics task is the most important calculation Joint forces Q that induce motion of the end-effector can be
procedure for control of the pedipulators. Analytical methods mapped to joint forces and torques for each calculation step k
cannot be simply used for the tasks due to complex mechanical with usage of (3) by application of manipulator Jacobian
structure, existence of redundancies and kinematic constraints transpose 𝐽𝐽(𝑞𝑞〈𝑘𝑘〉)𝑇𝑇 . The mapping is never singular as it can be
that arise in these closed kinematic chains. Numerical in the case of velocity kinematics using Jacobian approach,
differential kinematics with usage of manipulator Jacobian is therefore it can be efficiently used for numerical solution of
an effective approach for calculation of inverse kinematic. complex inverse kinematics problems (Corke 2011).
Depending on analysed structure, different techniques can be
utilized, including Jacobian transpose, pseudoinverse or 𝑄𝑄〈𝑘𝑘〉 = 𝐽𝐽(𝑞𝑞〈𝑘𝑘〉)𝑇𝑇 ∙ 𝐺𝐺𝐸𝐸 〈𝑘𝑘〉 (3)
damped least squares method (Buss 2009). For the closed
kinematic chain of the pedipulator structure, additional By assumption that the virtual robot does not have joint
conditions had to be applied and a dedicated algorithm for motors, but only viscous dampers that impose proportionality
numerical calculations was developed. of applied forces to joint velocity, it is possible to express
Calculation of manipulator Jacobian matrices was done on the generalized joint velocity by (4) and for a discrete time-update
basis of D-H transformation matrices, with use of a method of joint coordinates by (5), where B is a joint damping
shown in (Wu & Young 1990). Next, a numerical algorithm coefficient and αC is a gain selected according to desired
was used that on the basis of Virtual Work principle, iteratively calculation convergence speed.
minimizes error between initial and goal poses of the end-
effector. This method, whilst valid for dynamic modeling is 𝑞𝑞̇ 〈𝑘𝑘〉 = 𝑄𝑄〈𝑘𝑘〉/𝐵𝐵, (4)
useful in general inverse kinematics (Corke 2011). In Fig. 3, a 𝑞𝑞〈𝑘𝑘 + 1〉 = 𝛼𝛼𝐶𝐶 ∙ 𝑞𝑞̇ 〈𝑘𝑘〉 + 𝑞𝑞〈𝑘𝑘〉, (5)
scheme of the 3-DOF (rear) manipulator, extracted from the
pedipulator structure is presented. The calculations are performed iteratively until the magnitudes
of generalized external forces 𝐺𝐺𝐸𝐸 that actuate the manipulator
to desired pose are sufficiently small. With usage of this
method, numerical determination of inverse kinematics can be
executed. The analyzed robot has less than 6-DOF, therefore
a mask vector M has to be applied that specifies controllable
degrees of freedom. In this case it is used to select
𝑝𝑝𝐸𝐸𝐸𝐸 , 𝑝𝑝𝐸𝐸𝐸𝐸 , 𝜙𝜙𝐸𝐸𝐸𝐸 coordinates on a planar.

𝑄𝑄〈𝑘𝑘〉 = 𝐽𝐽(𝑞𝑞〈𝑘𝑘〉)𝑇𝑇 ∙ 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑀𝑀) ∙ 𝐺𝐺𝐸𝐸 〈𝑘𝑘〉. (6)

Finally, after substitution of (4), (5) and (6), discrete update of

joint coordinates for an under-actuated robot is given by (7).

𝐽𝐽(𝑞𝑞〈𝑘𝑘〉)𝑇𝑇 ∙𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑀𝑀)∙𝐺𝐺𝐸𝐸 〈𝑘𝑘〉

𝑞𝑞〈𝑘𝑘 + 1〉 = 𝛼𝛼𝐶𝐶 ∙ + 𝑞𝑞〈𝑘𝑘〉. (7)
𝐵𝐵

The method specified above is used for numerical inverse

Fig. 3. Scheme of the 3-DOF manipulator inverse kinematics kinematics calculations of the 3-DOF and also 2-DOF
– numerical solution. ξΔ - end-effector pose difference, manipulators, subdivided from the robot pedipulator structure.
ξE (pEx , pEy , ϕEz ) – end-effector current pose It is integrated as a part of the calculation procedure, described
in the trajectory calculation algorithm and plays an important
ξ∗E (pEx
∗
, p∗Ey , ϕ∗Ez ) – end-effector desired pose.
role in control system of the robot.
Principle of this method relies on a so called “special spring”
between current and desired pose. This spring is a set of 4. TRAJECTORY CALCULATION ALGORITHM FOR
generalized external forces 𝐺𝐺𝐸𝐸 that are able to change position TRANSFORMATION OF PEDIPULATORS
and orientation of the end-effector towards the desired pose. It
For trajectory planning, a dedicated algorithm was developed
is proportional to the difference between poses, where
that is composed of different methods utilized iteratively,
proportionality is denoted by a constant 𝛾𝛾𝑃𝑃 as stated in (1). numerical inverse kinematics based on Jacobian transpose and
Current pose at a given calculation step ξE 〈k〉, is computed pseudo-inverse, analytical inverse kinematics, forward
with forward kinematics (2), where q〈k〉 represents current kinematics, sets of geometric conditions and custom rules.
estimate of inverse kinematics solution and TE,0 is a
transformation matrix from end-effector to base coordinate Pedipulator poses dedicated for particular pipe sizes were
system. determined from the 3D CAD model of the robot. With usage
of this data, all angular positions of pedipulators joints were
GE ∝ Δ(ξE , ξ∗E ) → GE = γP ∙ Δ(ξE , ξ∗E ), (1) saved and used as initial and final poses for trajectory
generation. The second step was to generate 5th order
𝜉𝜉𝐸𝐸 〈𝑘𝑘〉 = 𝑇𝑇𝐸𝐸,0 (𝑞𝑞〈𝑘𝑘〉), (2) polynomial interpolated joint space trajectories for the front

134
IFAC SYROCO 2018
Budapest, Hungary, August 27-30, 2018 Michał Ciszewski et al. / IFAC PapersOnLine 51-22 (2018) 132–137 135

manipulator between initial and final poses. Next, inverse points are visualized in range from initial to final pose. The
kinematics problem was solved with usage of Jacobian visualization was prepared in MATLAB software.
transpose for the rear manipulator to match its position with
end-effector of the front manipulator. In case of this pipe
inspection robot, calculation of the pedipulator analytical
inverse kinematics problem gives 8 solutions for the entire
structure. Therefore, additional conditions were used for the
planar manipulators with limitation of joint rotation angles.
The final step was to eliminate oscillations of the manipulator,
caused by transition through singular positions. Finally, joint
trajectories of both 2-DOF and 3-DOF planar manipulators
were merged and only actuated joints were selected to perform
reconfiguration of the pedipulators. The calculations were
performed in MATLAB software with addition of the Robotics
Toolbox (Corke 2011). As a result, smooth trajectories for
transformation of both robot pedipulators were obtained. Due
to the fact that each pedipulator mechanism is a closed
kinematics chain of specific structure, it is sufficient to use
three motors to set particular pose of one track drive module.
Calculation workflow is depicted in Fig. 4. The scheme
constists of major steps in the calculations procedure. In some
cases, to achieve valid results, it is necessary to perform
several iterations, with use of custom, structure specific rules.

Fig. 5. Pedipulator transformation from the neutral pose to

a Ø210 mm horizontal pipe: drives positions, velocities and
accelerations.

(a)
Fig. 4. Calculation steps in pedipulators transformation
calculation algorithm.
On the basis of the previously presented mathematical models,
implemented in the original algorithm, joint trajectories for
different transformations were obtained. Transformation time
was assumed to be 5 seconds for all cases to ensure easier
comparison of the results. The results for transformation from
neutral pose to horizontal pipes with diameter Ø210 mm are
shown in Fig. 5. It can be noted that results of the
transformations are smooth due to utilization of 5 th order
polynomial interpolation and application of custom rules for (b)
trajectory generation task. The presented approach is effective
for solving inverse kinematics task for the closed kinematic Fig. 6. Graphical validation of pedipulators transformation
chain of the robot. trajectories from the neutral pose to a Ø210 mm horizontal
pipe: (a) initial pose; (b) final pose.
Graphical validation of the trajectories on a pedipulator model
is shown in Fig. 6. All characteristic points of the mechanical It can be observed that mathematical modeling of the robot can
structure were marked with cylinders, representing rotary be used for control of the pedipulators to adapt to different
joints. Traces of trajectories of all joints and track contact sizes and shapes of pipelines.

135
IFAC SYROCO 2018
Budapest, Hungary, August 27-30, 2018 Michał Ciszewski et al. / IFAC PapersOnLine 51-22 (2018) 132–137
136

5. CO-SIMULATION IN MATLAB AND V-REP To check whether the prototype would be capable of driving
SOFTWARE on rough surfaces, a special simulation environment was
prepared. The motion on a randomly generated rough terrain,
Co-simulations of the robot control in V-REP and MATLAB as depicted in Fig. 7b gave stable simulation results.
environments were prepared to analyze and verify
mathematical models of the robot, especially trajectory Operation of the robot model in the V-REP environment was
calculation for adaptation of pedipulators to different work compared to the theoretical results of pedipulators trajectory
environments. V-REP is a multi purpose robotic simulation planning, obtained in MATLAB. In Fig. 8, comparison of
software that allows integration of control algorithms or pedipulator transformation results for the structure are plotted
connection with external applications to run the models (Spica for the complex 2-phase adaptation between neutral pose and
et al. 2014). Robot model creation process was thoroughly a pose for motion on horizontal surfaces via an intermediate,
described in (Ciszewski et al. 2016). Co-simulation of the empirically determined pose. The absolute position error
robot model required creation of a MATLAB/Simulik between those two models do not exceed 1.5 degrees. It should
schematic diagram that allowed integration of the previously be noted that the theoretical results obtained in MATLAB do
described trajectory planning algorithm with hardware support not take into consideration dynamic effects as forces in joints
of a physical joystick used for control of the robot model, as that are actuated or free in the V-REP simulation, as in real
well as synchronous communication with the V-REP prototype. Moreover, in the V-REP model, the Bullet
simulator. Initial and final poses for transformation of the simulation engine, used for the simulations, adds some
pedipulators were chosen by the operator, based on knowledge elasticity and damping to revolute joints. Therefore, the results
about pipe gemetry. In real operating conditions, these can be viewed as valid, since very precise positioning is not
decisions can be made based on camera video signal. crucial for proper operation of the prototype.
The simulations involved tests in horizontal pipes with
different diameters and obstructions such as bends and
reducers. In Fig. 7a, the robot after negotiation of a bend is
shown. Reconfiguration of the pedipulators is controlled with
usage of the control algorithm developed in MATLAB,
whereas, track drives velocities are adjusted in real time by the
operator by the hardware joystick, handled by
MATLAB/Simulink synchronous simulation.

Fig. 8. Comparison of drives positions during pedipulators

transformation - MATLAB numerical model and V-REP
simulator: from the neutral pose to a pose for motion on
horizontal surfaces and rough terrain.

6. PROTOTYPE OF THE PIPE INSPECTION ROBOT

(a) A prototype of the tracked mobile robot for pipeline inspection
is shown in Fig. 9. Subsystems of the robot were integrated
with electronics and a control system, designed for
compatibility with the trajectory planning algorithm, remote
operation by a joystick and transmission of video signal to
operator by a tether cable. Prototype tests were conducted in
different environments. The robot during a test of chassis
adaptation is shown in Fig. 9. Transformation of pedipulators
is coherent with simulation results. Different scenarios of
chassis adaptation were checked, including motion in
horizontal pipes with bends and reducers, based on trajectories
obtained from the original algorithm.
(b)
In Fig. 10 the robot is shown during visual inspection of a
Fig. 7. Co-simulation of the robot motion in V-REP and horizontal pipe. Watertightness of robot components was
MATLAB software: (a) horizontal Ø315 mm pipe; (b) successfully verified and the test proved that adaptation of the
horizontal surfaces and rough terrain. tracks to move on pipe walls provides enhanced traction,
especially in pipes filled with sediments in lower sections.

136
IFAC SYROCO 2018
Budapest, Hungary, August 27-30, 2018 Michał Ciszewski et al. / IFAC PapersOnLine 51-22 (2018) 132–137 137

REFERENCES
ABE Group, 2017. Kamery do inspekcji rurociągów i
kanalizacji. Available at: http://abe-group.eu/kamery-
inspekcyjne/ [Accessed September 26, 2017].
Buratowski, T., Uhl, T. & Żylski, W., 2003. The comparison
of parallel and serial-parallel structures of mobile robot
pointer 2DX state emulator. IFAC Proceedings
Volumes, 36(17), pp.295–300.
Buss, S.R.S., 2009. Introduction to inverse kinematics with
Fig. 9. Prototype of the robot – adaptation test of the jacobian transpose, pseudoinverse and damped least
pedipulators with usage of the algorithm squares methods. University of California, San Diego,
Typeset manuscript, 132(4), pp.1–19.
Ciszewski, M. et al., 2013. Mobile inspection robot. In
Applied Mechanics and Materials. pp. 385–392.
Ciszewski, M. et al., 2016. Modeling and simulation of a
tracked mobile inspection robot in MATLAB and V-
REP software. In K. Tchoń & C. Zieliński, eds. Postępy
robotyki. Warszawa: Prace Naukowe Politechniki
Warszawskiej. Elektronika, pp. 135–144.
Ciszewski, M. et al., 2014. Virtual Prototyping, Design and
Analysis of an in-Pipe Inspection Mobile Robot.
Journal of Theoretical and Applied Mechanics, 52(2),
pp.417–429.
Corke, P., 2011. Robotics, Vision and Control: Fundamental
Fig. 10. Prototype of the robot – adaptation test of the Algorithms in MATLAB, Springer Science & Business
pedipulators with usage of the algorithm Media.
Dulęba, I. & Opałka, M., 2013. A comparison of Jacobian-
7. CONCLUSIONS based methods of inverse kinematics for serial robot
Application of mathematical tools and computer simulations manipulators. International Journal of Applied
proved to be effective and provided initial testing platform of Mathematics and Computer Science, 23(2), pp.373–
control system designed for the robot’s pedipulators. The 382.
modeling approach with use of differential kinematics and Inuktun, 2017. Inuktun | The Multi-Mission Modular
iterative solution of inverse kinematics problem can be Robotics Company. Available at:
successfully used for control of closed kinematic chains, but it http://inuktun.com/en/products/ [Accessed September
is necessary to enhance the model with custom structure- 26, 2017].
dependent rules. Results of the experiments with the prototype NEOVISION s.r.o., Jetty, Cleaning and inspectional robot for
lead to a conclusion that it is advantageous to utilize an active air-induction ducting. Available at:
chassis with adaptation capabilities to apply one robot for http://www.neovision.cz/sols/jetty.html [Accessed
different types of environments with restricted access. March 31, 2016].
Mathematical modeling and simulations of mechatronic Spica, R. et al., 2014. Interfacing Matlab / Simulink with V-
systems have been developed in the Department of Robotics REP for an Easy Development of Sensor-Based
and Mechatronics, AGH for many years, however in initial Control Algorithms for Robotic Platforms. In 2014
phase they relied only on numerical calculations performed in IEEE Int. Conf. on Robotics and Automation Workshop
MATLAB/Simulink (Buratowski et al. 2003). Application of on MATLAB/Simulink for Robotics Education.
modern simulation software such as V-REP gives much better Brisbane.
possibilities to integrate mathematical modeling, implemented Wang, J., Li, Y. & Zhao, X., 2010. Inverse kinematics and
in MATLAB and visualize their applicability on a virtual control of a 7-DOF redundant manipulator based on the
model to create effective co-simulations, enhanced by closed-loop Algorithm. International Journal of
multibody simulation, integrated motor controllers, emulation Advanced Robotic Systems, 7(4), pp.1–9.
of sensors and environmental interaction. Wu, C.H. & Young, K.Y., 1990. An efficient solution of a
differential inverse kinematics problem for wrist-
The future works would include elaboration of an active partitioned robots. IEEE Transactions on Robotics and
adaptation system with regulation of extension forces, based Automation, 6(1), pp.117–123.
on measurements from current drain sensors for the Zhang, Y. & Yan, G., 2007. In-pipe inspection robot with
servomotors and the tracks. This system would provide active pipe-diameter adaptability and automatic tractive
improved safety during motion in vertical pipelines with force adjusting. Mechanism and Machine Theory,
sediments, obstructions and other disturbances of pipe 42(12), pp.1618–1631.
geometry by automatic adjustment of clamp force.

137