TYPE Technology and Code

PUBLISHED 21 February 2023

DOI 10.3389/frobt.2023.1120658

Research on automatic
OPEN ACCESS emergency steering collision
EDITED BY
Xin Xia,
University of California, Los Angeles,
avoidance and stability control of
United States

REVIEWED BY
intelligent driving vehicle
Wei Liu,
Purdue University, United States
Weihua Li,
Zhaoyong Liu 1,2, Gaobo Wen 1, Wudong Liu 1, TanXiaoqiang Tan 1
Harbin Institute of Technology, Weihai, and Guangqiang Wu 1*
China
1
School of Automotive Studies, Tongji University, Shanghai, China, 2Global Technology Co., Ltd, Nantong,
*CORRESPONDENCE
China
Guangqiang Wu,
[email protected]

SPECIALTY SECTION
This article was submitted
to Robotic Control Systems, In view of the need for emergency steering to avoid collision when the vehicle is in
a section of the journal a dangerous scene, and the stability control of the vehicle during collision
Frontiers in Robotics and AI
avoidance. This paper proposes a planning and control framework. A path
RECEIVED 10 December 2022
planner considering the kinematics and dynamics of the vehicle system is used
ACCEPTED 06 February 2023
PUBLISHED 21 February 2023 to formulate the safe driving path under emergency conditions. LQR lateral
control algorithm is designed to calculate the output steering wheel angle. On
CITATION
Liu Z, Wen G, Liu W, Tan T and Wu G this basis, adaptive MPC control algorithm and four-wheel braking force
(2023), Research on automatic distribution control algorithm are designed to achieve coordinated control of
emergency steering collision avoidance
vehicle driving stability and collision avoidance safety. The simulation results show
and stability control of intelligent
driving vehicle. that the proposed algorithm can complete the steering collision avoidance task
Front. Robot. AI 10:1120658. quickly and stably.
doi: 10.3389/frobt.2023.1120658

COPYRIGHT
KEYWORDS
© 2023 Liu, Wen, Liu, Tan and Wu. This is
an open-access article distributed under trajectory planning, stability control, LQR lateral control, emergency collision avoidance,
the terms of the Creative Commons direct yaw moment control
Attribution License (CC BY). The use,
distribution or reproduction in other
forums is permitted, provided the original
author(s) and the copyright owner(s) are Introduction
credited and that the original publication
in this journal is cited, in accordance with
The Advanced Driver Assistance Systems (ADAS) are effective in reducing crashes. Most
accepted academic practice. No use,
distribution or reproduction is permitted ADAS systems have one thing in common, that is, they all affect the longitudinal control of
which does not comply with these terms. the vehicle to avoid the collision (Borrello et al., 2020; Rabhi et al., 2021; Zha et al., 2021;
Hang et al., 2022; Wu et al., 2023).
Although, there are certain situations where a collision cannot be avoided by braking but only
by steering operations. However, lots of studies show that in the case of an impending rear-end
collision, many drivers tend to only brake rather than try to avoid obstacles by steering (Schieben
et al., 2014). There are different reasons for this behavior. Firstly, it is an instinctive reaction to
stop in order to reduce the impact of an impending collision. Secondly, steering is more complex
than braking, so it requires the driver to have a higher awareness of the situation and a higher
driving ability. Therefore, Automatic Emergency Steering System (AES) is of great significance.
For vehicle trajectory planning, Yang proposed a dynamic planning method for vehicle
collaborative trajectory planning under the scenario of forced lane change, which aims to
provide suggested lane change distance and reference trajectory for each autonomous vehicle
in a coordinated manner (Yang et al., 2022). A dynamic programming way is established to
determine the suggested distance of all vehicles and the non-convex quadratic constraint is
applied to characterize the trajectory determination problem. Considering driver comfort
and collision risk, Li et al. (2022) proposed a human-like motion planning strategy based on

Frontiers in Robotics and AI 01 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

probabilistic prediction under dynamic environment. They realized planning and control plays an important role in avoiding collisions
path generation based on quintic polynomials, and optimized the while stabilizing autonomous vehicles in extreme scenarios. Liu et al.
target trajectory by using cost functions with four indexes including (2017) proposed a method to establish the stability criterion of vehicle
safety, consistency, smoothness and distance from local path to yaw based on the phase plane method of sideslip-yaw rate, which solved
global path. Cheng et al. (2022) proposed a deep reinforcement the problem of judging the type of vehicle stability region under different
learning method based on time difference to solve the longitudinal driving conditions, and provided a theoretical basis for the intervention
trajectory planning of autonomous vehicles at signal-controlled algorithm of stability control system. Zhang et al. (2017) considered the
intersections. Xiao et al. (2021) [10] proposed the control barrier influence of tire slip and actuator torque saturation on driving and
function method for critical safety control and developed a real-time braking, and designed a dynamic controller to overcome integral
control framework that combines the optimal trajectory generated saturation by using a conditional integrator to ensure accurate
with the computational efficiency method that provides safety tracking of the required signals under the influence of tire force and
assurance. actuator constraints. Vehicle state and parameter estimation is an
Vehicle collision risk assessment is the key to trigger AES. For important part of vehicle dynamic control. Liu et al. (2021) proposed
vehicle risk assessment, Li et al., 2021 proposed a multi-scene collision a new estimation method of vehicle side-slip angle based on kinematic
avoidance decision algorithm for autonomous vehicles, and used the model, which integrated the information of Global Navigation Satellite
situation assessment module based on conditional random field to System (GNSS) and inertial Measurement Unit (IMU). Xia et al. (2018)
assess the risk level of the surrounding traffic participants. Based on the proposed a method to estimate the attitude and lateral velocity of an
situation assessment module, collision avoidance strategies with driving autonomous vehicle with the assistance of vehicle dynamics using a six-
style preferences (such as aggressive or conservative) are proposed to degree-of-freedom IMU. Liu et al. (2018) proposed a method based on
meet the needs of different drivers. Cui et al. (2021) proposed a layered kinematics model that integrates intelligent vehicle sensors to estimate
framework of manned or autonomous vehicles for collision avoidance sideslip angle, aiming at the problem that the non-linear characteristics
in emergency situations. They adopted finite-state machine (FSM) and parameter uncertainties of vehicles make it difficult for the method
technology to determine appropriate strategies for collision based on dynamic model to estimate the sideslip angle of vehicles under
avoidance, and established a collision risk model, taking into harsh working conditions. Xiong et al. (2020) proposed a new automatic
account vehicle risks around overlapping areas, road attachment vehicle sideslip angle and attitude estimation method based on IMU for
risks and vehicle stability performance. Gilbert et al. (2021) low sampling rate GNSS speed and position parallel adaptive Kalman
proposed a decision-making system that selects the lightest collision filters.
when vehicles are confronted with inevitable collisions on the highway. The existing steering collision avoidance system often only
They applied the multi-attribute decision-making method to judge the considers the safety risk of collision avoidance, but does not
severity of the collision. For the autonomous lane change decision of consider the impact of dynamic factors on stability. At the same
trucks, Chen et al. (2020) proposed a lane change decision model based time, stability has a certain impact on tracking control accuracy,
on support vector machine. Ren and Wu, 2020 proposed a fusion which should also be considered. Therefore, the contribution of this
architecture of decision planning under dynamic Environment And paper is to propose a real-time emergency steering collision
Used Back Propagation Neural Network (BPNN) to predict the lane avoidance and stability control method, and design a simulation
change of vehicles around the block. experiment based on the influence of stability control on tracking
The AES control layer will track the trajectory planned by the accuracy and other factors.
decision layer. For the vehicle trajectory tracking, Ge et al., 2022 rely on As shown in Figure 1, the lateral path planning and path tracking
the precise model for the traditional MPC. When the autonomous control considering motion stability designed in this paper are parts of
vehicle encounters external interference and perturbation, the steady- the coordinated control architecture of vehicle driving stability and
state non-offset tracking cannot be realized, and the MPC solver is collision avoidance safety, which can realize automatic collision
biased to solve the coupling control problem. Li et al. (2021) studied that avoidance control and ensure the vehicle’s security and stability.
under extreme driving conditions, the coupling between the Based on perception and state estimation information, the
longitudinal and transverse motion of the vehicle becomes framework judges driving safety and collision risk and makes
significant due to the highly non-linear force of the tire, which decisions based on TTC (time to collision), and uses dynamic
affects the stability of the vehicle. They proposed a model prediction programming and quadratic programming methods to plan paths
controller of electric vehicle driven by four-wheel independent motor, and determine collision-free paths. Then, according to the
in which changes in the longitudinal velocity are regarded as characteristics of stability control in emergency collision avoidance
interference in the vehicle dynamics model. Then, the additional scenarios, an adaptive MPC control algorithm is designed. Finally, the
torque generated by the model-based controller with the multi- obtained steering wheel angle and four-wheel braking force are applied
objective design is considered for balance. Tork et al. (2021) to the actual vehicle.
proposed an independent model control based on neural network
for path tracking control. The control scheme utilized the input of
steering Angle and torque to realize cooperative control of transverse Vehicle dynamics model
and longitudinal motion.
In summary, collision avoidance and stabilization are the two critical 3-DOF vehicle dynamics model
issues when an autonomous vehicle in an emergency situation, which
usually occurs in a short time and requires large actuator inputs, as well We rationally simplified the vehicle model (Wu G. 2021) to
as a highly non-linear response. Real-time vehicle decision-making, obtain a three-degree-of-freedom(3-DOF) vehicle dynamics model,

Frontiers in Robotics and AI 02 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

FIGURE 1
Coordinated control for vehicle driving stability and collision avoidance safety.

as shown in Figure 2, which mainly includes vehicle longitudinal, F Yi
kf βf i
1, 2

(2)
lateral and yaw motions, where δ f is the front wheel angle, and FXi is F Yi
kr βr i
3, 4
the wheel braking force, FYi is the wheel lateral force, vx is the
longitudinal speed, vy is the lateral speed, v is the speed of the Where kf and kr are the cornering stiffnesses of the front and rear
vehicle, β is the sideslip angle, lf and lr are the distance of the center axles, respectively. Front and rear wheel sideslip angles, vehicle
of mass and the front and rear axles, ls is the wheelbase, βf is the sideslip angle and their derivatives are:
sideslip angle of the front wheel, Td is the sum of the yaw moment of ⎪
⎧
⎪ lf r
⎪
⎪ β
β+ − δf
the vehicle, r is the yaw rate of the vehicle. ⎨ f vx
⎪ (3)
The differential equations of motion of the vehicle: ⎪
⎪ lr r
⎪
⎩ βr
β −
vx
⎪
⎧
⎨
F xi
mv_ x − r × v y 
⎪ vy
⎪
⎧
⎪ β


F yi
mv_ y + r × v x  ⎪
⎨ vx
⎪
⎪
⎩l F +F −l F +F +l F +F −l F +F
T ⎪ (4)
f  y1 y2  r  y3 y4  s ( x1 x3 ) s ( x2 x4 ) ⎪
⎪ v_ v − v_ v
d
⎩ β_
y x 2 x y
(1) vx

So far, the establishment of the 3-DOF model of the vehicle

considering the lateral, longitudinal and yaw motions has been
completed, and this model will be used to describe the basic Collision avoidance trajectory planning
characteristics of the vehicle during motion.
The goal of trajectory planning is to generate a smooth enough
curve to change the position of the vehicle under the premise of
Linear tire model ensuring the safety of the vehicle. The smoothness is to ensure that
the vehicle can track along the trajectory. The trajectory planning
In the case of a small vehicle front wheel angle, the relationship module will receive the environment information including vehicle
between the wheel lateral force and the sideslip angle of this wheel can location information and road information. A planned trajectory is
be approximately regarded as a linear relationship (Wu G. 2021), thus: transmitted to the vehicle motion control module as shown in Figure 3.

Frontiers in Robotics and AI 03 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

correspond to the cost weights generated by the first, second, and

third derivatives of the path:
T T T
Cpsmooth
Wpsmooth1 y′ y, +Wpsmooth2 y″ y″ + Wpsmooth3 y″′ y‴
(5)

The reference line cost is Cpref , and Wpref represents the

corresponding weight:

Cpref
Wpref yT y (6)

The obstacle cost is Cpcollision , Wpcollision represents the

corresponding weight, and d represents the distance between the
obstacle and the vehicle, where Wpcollision is a rather large value.
⎪
⎧
⎪ 0 , if d ≥ 4
⎪
⎪
⎨ 1000
Cpcollision
⎪ , if 3 < d < 4 (7)
⎪
⎪ d
⎪
⎩
Wpcollision , if d ≤ 3

Combining the above three formulas, the total planning cost

Cpnode of each discrete point can be obtained, and its value is equal
to the sum of the above three costs:
Cpnode
Cpsmooth + Cpref + Cpcollision (8)

The planned trajectory needs that the curvature is continuous.

Δx is the sampling interval in the direction of the road centerline, yi ′
FIGURE 2 is the first derivative of y to x at the i-th sampling point, yi ″ is the
3-DOF vehicle dynamics model. second derivative of y to x at the i-th sampling point:
1 1 yi+1 ″ − yi ″
yi+1 ′
yi ′ + yi ″Δx + yi ″Δx2 + Δx2 (9)
2 2 Δx
Usually trajectory planning can be decomposed into speed planning
The solution of the planning is first to obtain the initial solution
and path planning. Since the change of speed is not considered in the
by dynamic programming, and then to obtain the final result by
planning process of steering collision avoidance, we assume that the
secondary programming, as shown in Figure 4. In order to speed up
longitudinal speed is constant in the planning process, and only the path
the calculation of the quadratic programming problem, the result of
planning is designed here.
the dynamic programming is regarded as a rough solution and a
feasible space is opened up. ymax(j) , ymin(j) is the maximum and
minimum value of the feasible space, y′min(j)
′
, ymax(j)
is the value range of the
Path planning path restricted by the road, y obs is the position corresponding to
the obstacle car, width/2 is the width of the obstacle car:
Using numerical methods, the path planning problem in
discrete space can be abstracted as a quadratic programming ⎪
⎧
⎪ ′
⎨ ymax(j)
min ymax(j) , y obs − width 2, if ydp path > y obs
problem, and the construction of this problem mainly includes
⎪
⎪
two parts: the design cost function and the determination of the ⎩ ymin(j)
max y′min(j) , y obs + width 2, others
constraints of the problem. In designing the cost function, we (10)
need to consider the requirements of smoothness, not deviating
from the road centerline and being away from obstacles. At the After a lot of tuning, the parameters of the
same time, in order to accelerate the solution, we first use planning algorithm are finally selected
dynamic programming to open up feasible space and therefore Wpsmooth1
15; Wpsmooth2
20000; Wpsmooth3
5000; Wpref
15.
determine the constraints of the planning problem. After the
problem is constructed, we use the iterative method to solve the
quadratic programming problem. Quadratic programming problem solving
The cost function of path planning can be divided into
smoothing cost, reference line cost and obstacle cost. The road is The advantage of Dynamic Programming (DP) is to decompose
discretized along the centerline and its perpendicular direction, x is each column of the discrete space into a sub-problem and solve the
the coordinate of the road centerline, and y is the coordinate of the optimal path from the last column through the inverse method. On
point which is perpendicular to the road. the basis of discretization of the solution space, the initial path can be
The smoothing cost Cpsmooth is divided into three parts, calculated by using dynamic programming. According to this path, a
Wpsmooth1 , Wpsmooth3 , and Wpsmooth3 , and their meanings preliminary decision can be made on the path planning problem to

Frontiers in Robotics and AI 04 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

FIGURE 3
Trajectory planning during the steering.

FIGURE 4
Path planning feasible space(DP).

reduce the search range of the quadratic feasible space, as shown in emergency collision avoidance with the stability of the vehicle body, as
Figure 4. The advantage of the iterative method for solving quadratic shown in Figure 5. The structure of the system can be mainly divided
programming problems is that it can balance the solution time and into three parts: TTC risk assessment, LQR lateral control and adaptive
effect. This article does not focus on this aspect, so directly call the Model Predictive Control (MPC) stability control.
quadratic planner solution function in MATLAB. First, according to the collision risk assessment module, the collision
time TTC is calculated according to the state of the vehicle and the
environment perception information to judge the safety of the current
Control strategy of collision avoidance vehicle. TTC refers to the time it would take for a collision to occur at the
system prevailing speeds, distances, and trajectories associated with the driver’s
vehicle and the closest lead vehicle. TTC can be kinematically defined as
In this paper, a path-following control system considering motion the range between a following and lead vehicle divided by the relative
stability is proposed. Its purpose is to judge the risk of collision when an velocity between these vehicles. Hence, TTC provides a measure of crash
obstacle appears in front of the vehicle, and automatically implement risk or the time before impact if prevailing conditions continue (Coelingh

Frontiers in Robotics and AI 05 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

FIGURE 5
Vehicle emergency steering and collision avoidance stability control.

change of heading angle error e_h . The following formula is the state space
equation of the system:
0 1 0 0
⎡⎢⎢⎢ ⎤⎥⎥⎥
e_y ⎢⎢⎢ 1 1 ⎥ e
⎡⎢⎢⎢ ë ⎤⎥⎥⎥ ⎢⎢⎢⎢ 0
1
(kf + kr  − k f + k r  lf kf − lr kr  ⎥⎥⎥⎥⎥⎡⎢⎢ y ⎤⎥⎥
⎢⎢ y ⎥⎥
⎢⎢⎢
mu
m mu ⎥⎥⎥⎢⎢⎢ e_y ⎥⎥⎥
⎢⎢⎣ e_ ⎥⎥⎦ ⎢⎢ ⎥⎥⎥⎢⎢⎢ ⎥⎥⎥
⎢⎢⎢ 0 0 0 1 ⎥⎥⎥⎣ eh ⎦
h
⎢⎢⎢ ⎥⎥⎥
ëh ⎢⎣ 1 1 1 2 2
⎥⎦ e_h
0 l f k f − l r k r  − l f k f − l r k r  lf kf + lr kr 
I z vx Iz I z vx
0 0
⎡⎢⎢⎢ ⎤⎥ ⎡⎢⎢⎢ ⎤⎥⎥⎥
⎢⎢⎢ kf ⎥⎥⎥⎥⎥ ⎢⎢ 1 ⎥
⎢⎢⎢ − ⎥⎥⎥ ⎢⎢⎢  akf − bkr  − vx ⎥⎥⎥⎥
⎢⎢⎢ m ⎥⎥⎥ ⎢⎢⎢ mvx ⎥⎥⎥
+ ⎢⎢⎢⎢ ⎥⎥⎥δf + ⎢⎢⎢
⎢⎢⎢ ⎥⎥⎥⎥δ_f (11)
⎢⎢⎢ 0 ⎥⎥⎥⎥ ⎢ 0 ⎥⎥⎥
⎢⎢⎢ ⎥⎥⎥ ⎢⎢⎢⎢ ⎥⎥⎥⎥
⎢⎣ akf ⎥⎦ ⎢⎣ 1 ⎥⎦
−  a2 k f + b 2 k r 
Iz I z vx

Where:
FIGURE 6 0 1 0 0
Four-wheel braking force distribution. ⎢
⎡
⎢
⎢ ⎤⎥⎥⎥
⎢
⎢
⎢ 1 1 ⎥⎥⎥
⎢
⎢
⎢ 0 1
(kf + kr  − kf + kr  lf kf − lr kr  ⎥⎥⎥
⎢
⎢
⎢ mvx ⎥⎥⎥
⎢ m mvx ⎥⎥⎥
A
⎢
⎢
⎢
⎢ ⎥⎥⎥,
⎢
⎢
⎢ 0 0 0 1 ⎥⎥⎥
et al., 2010; Han et al., 2014). We have carried out related research in this ⎢
⎢
⎢ ⎥⎥⎥
⎢
part, but it is not the focus of this article. Considering the complex and ⎢
⎢
⎢
⎣ 1 1 1 ⎥⎥⎦⎥⎥
0 lf kf − lr kr  − lf kf − lr kr  lf 2 kf + lr 2 kr 
changeable traffic conditions, as well as the possible instability of the I z vx Iz I z vx
vehicle caused by the large action of the actuator, this paper designs an 0
⎢
⎡
⎢
⎢ ⎤⎥⎥⎥
adaptive MPC module to control the yaw moment of the vehicle to ⎢
⎢
⎢ kf ⎥⎥⎥⎥
⎢
⎢
⎢ − ⎥⎥
ensure the stability of the vehicle. The module judges whether to ⎢
⎢
⎢ m ⎥⎥⎥⎥⎥
⎢
⎢
⎢ ⎥⎥⎥,
intervene. In addition, the linear quadratic regulator (LQR) controller B
⎢⎢
⎢
is proposed in this paper to calculate the output signal steering angle δ
⎢
⎢
⎢
⎢ 0 ⎥⎥⎥⎥⎥
⎢
⎢
⎢ ⎥⎥⎥
⎢
⎢ ⎥
according to the lateral error eh and heading angle error ey , so that the ⎣ akf ⎥⎥⎦
⎢
−
vehicle can always track the road centerline. Finally, in terms of braking Iz
force control and distribution, the braking force distribution and control 0
⎢
⎡
⎢
⎢ ⎤⎥⎥⎥
module will calculate the braking forces FX1 、FX2 、FX3 、FX4 ⎢
⎢
⎢ akf − bkr ⎥⎥⎥
⎢
⎢
⎢ − δ ⎥⎥⎥
according to the expected deceleration and the expected additional ⎢
⎢
⎢ f ⎥⎥⎥
⎢
⎢ mvx ⎥⎥⎥
yaw moment of the vehicle, and From this, the braking pressures C
⎢
⎢
⎢
⎢
⎢
⎥⎥⎥
⎥⎥⎥
⎢
⎢
⎢ 0 ⎥⎥⎥
PX1 、PX2 、PX3 、PX4 of each wheel cylinder are further obtained. ⎢
⎢
⎢
⎢
⎢
⎢ ⎥⎥⎥⎥
⎢
⎣ a 2
k f + b2
k r ⎦ ⎥
I z vx
LQR lateral control (12)

The main purpose of lateral control is to control the lateral error Eq.11 can be expressed as:
within a certain range. As a result we can get a better track of the desired
e_rr
Aerr + Bu + C θ_r (13)
path and the heading angle of the vehicle. The content of this section is
mainly based on the LQR to design the controller to track path. The state Error:
variables of the control system are four parameters, including: lateral error
T
ey , rate of change of lateral error e_y , and heading angle error eh , the rate of err
 ey e_y eh e_h  (14)

Frontiers in Robotics and AI 06 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

The control quantity is δ f . MPC is a feedback control strategy that discretizes the vehicle
From the above state equation, we can get the objective function dynamics equation, and sets the sampling time as Ts
5 × 10−4 s.
and corresponding constraints of the LQR controller: For time k, there is the following discrete equation:
1 tf x(k + 1)
Ax(k) + Bu u(k) + Bd d (k)
Min J
 (err (t )T Q err (t) + u(t )T R u(t)dt (21)
2 t0 (15) y(k)
Cx(k) + Du(k)
s.t. e_rr
Aerr (t ) + Bu Where,
Where [t0 , tf ] is the time domain, Q and R represent the weighted A
T s p Ac + I, Bu
Bcu T s , Bd
Bcd T s , C
C c , D
Dc
matrix of state and control quantity.
The optimal solution of this problem satisfies the following: x(k), u(k), y(k) represent the state, control input and output
of the system at time k, respectively. Assuming that the
J*
err (t )T P err (t ) (16) prediction time domain is ps , and the control time domain is
ms , the control quantity will not change when the time exceeds
The expression of P is:
the control time domain. Then the input and output predicted by
P_
P(t ) A(t ) + AT (t ) P(t ) − P(t )B(t )R−1 (t )BT (t )P(t ) + Q(t ) time k are:
(17)

yp (k + 1|k), yp (k + 2|k), ..., yp k + ps k (22)
The LQR controller is: 
u(k|k), u(k + 1|k), ..., uk + ps − 1k   (23)
uk
−K (t ) X (t ) (18)
The control goal is to track the target and reduce the tracking
Where K
−R−1 (t)BT (t) P(t) represents the controller gain.
error, that is,
r (k + 1), r (k + 2), ..., r k + ps  (24)
Adaptive model predictive control stability
control module At the same time, the control constraints and output constraints
of the system are set. Finally the optimization goal function that can
In order to trade off the calculation efficiency and calculation characterize the control performance of the system is proposed. It
accuracy, we assume that the longitudinal velocity vx in the Formula. needs to consider the cost of the expected tracking error and some
1 of the vehicle dynamics model remains unchanged. At this time, other performances, such as the control action as small as possible.
the three-degree-of-freedom model of the vehicle is simplified to The objective function is:
!
ps !
!
two-degree-of-freedom. At the same time, it is brought into the
J y(k), U (k)

i
1 !!Γ r (k + i) − y (k + i|k)!!!
! y p !2
Formula. 4, where the longitudinal velocity vx is a time-varying ms
model parameter, and the sideslip angle and the yaw rate are taken as +
i
1 Γ u (u(k + i|k) − u(k + i − 1|k))2 (25)
the state quantities, which can be finally simplified to obtain the
Γy is the weight of the output quantity, and Γy u is the weight of the
following state space equation:
control quantity increment. For time k, the open-loop optimization
l f 2 k f + l r 2 kr l f k f − l r k r lf kf problem is transformed into solving min(J(x(k), U(k), ms , ps )) for
⎡⎢⎢⎢ ⎤⎥⎥⎥ 1 ⎡⎢⎢⎢ − ⎤⎥⎥
r_
 _ 
⎢⎢⎢⎢
⎢⎢ vx I z Iz ⎥⎥⎥ r ⎢⎢⎢ ⎤⎥⎥⎥
⎡ ⎢ I z ⎥⎥⎥⎥⎥δ
⎢ the control variable U.
⎥⎥⎥⎥ β  + ⎢⎣ I z ⎥⎦T d + ⎢⎢⎢⎢ ⎥ f
β ⎢⎢⎣ kf lf − kr lr kf + kr v_x ⎥⎦ ⎣ kf ⎥⎥⎦ In order to improve the calculation efficiency, the
−1 − 0 −
mvx 2 mv x vx mu longitudinal velocity vx is assumed to be constant when
(19) calculating the state space equation, but in the process of
emergency collision avoidance, its longitudinal velocity vx is a
Where select x
[ r β ]T as the state quantity, and u
Td as the time variable. At this time, the internal model of MPC will also
control quantity. change with time, so an adaptive MPC solution method is
Eq. 9 will calculate the reference values of the vehicle’s yaw rate proposed. The longitudinal velocity and longitudinal
and sideslip angle. The deviation value is used as the index of vehicle acceleration output by the system are fed back to the MPC
lateral stability, and the larger the value, the greater the risk of lateral controller to update the internal model of the controller,
instability of the vehicle. which is beneficial to improve controller performance.
⎪
⎧
⎪ vx lf + lr  After a large number of parameter tuning and system
⎪
⎪ r ref
δf
⎪
⎪ 1 + Kv x 2 identification, the system can maintain the best performance
⎪
⎪
⎪
⎪ as much as possible, and the corresponding parameters of the
⎪
⎨ lr + mlf vx 2 kr lf + kf lr 
β
(20) adaptive MPC controller are selected: prediction time domain
⎪
⎪ ref
lf + lr 1 + Kvx 2 
⎪
⎪ ps = 10, control time domain ms = 5, and add hard constraints to
⎪
⎪
⎪
⎪ the control input, Take u min
4000Nm, u max
4000Nm,
⎪
⎪ m lf l r
⎪
⎩K
2 − Δu max
± 1000Nm, Γu
0.02, Γy
[ 1 2 ].
lf + lr  kr kf
So far, the parameter setting of the adaptive MPC controller is
rref , βref are the reference yaw rate and sideslip angle, completed, and the optimal additional yaw moment ΔM can be
respectively, and K is the stability factor. calculated to avoid vehicle instability.

Frontiers in Robotics and AI 07 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

FIGURE 7
Tire selection strategy.

FIGURE 8
Carsim simulation diagram.

Vehicle tire braking force distribution Fri , Ffi are the longitudinal forces of the front and rear tires,
strategy dri , dfi are the lateral distances from the front and rear tires to the
center of gravity of the vehicle. I represents the left or right side of the
The additional yaw moment Td required by the vehicle has been vehicle.
obtained from the MPC controller as shown in Figure 6: At present, the braking force distribution schemes for active braking
4 of vehicles can be roughly divided into two types, single-wheel braking
T d

i
1 (−1)i F xi Li (26) and multi-wheel braking. The braking force provided by the multi-wheel

Frontiers in Robotics and AI 08 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

FIGURE 9
(A) Ego vehicle speed (B) Obstacle avoidance trajectory.

FIGURE 10
(A) Variation of yaw rate with time (B) Variation of heading angle with time.

braking scheme is greater than that of the single-wheel braking scheme, wheels on the inner side of the steering wheel perform active braking to
but at the same time, the impact generated by the double-wheel braking correct the body condition.
scheme in the active braking process is also relatively large. In addition, Then, it is necessary to select the most effective wheel to generate
the additional yaw moment generated by the wheel braking scheme is also Td according to the actual situation. If the front wheels cannot
larger. Therefore, in the face of emergencies, a multi-wheel braking provide enough additional yaw moment, the remaining yaw
scheme with faster control speed and larger upper limit of additional moment can be generated by the training wheels. In order to
yaw moment is often adopted. design the wheel selection strategy, define the following formula:
During the turning, the effects of braking different wheels on the ƛ
ƛM · ƛγ , ƛM
sgn(T d ), ƛγ
sgn(r ) (27)
steering dynamic performance are different, and the single-wheel
control strategy will select different wheels for control under As shown in Figure 7, the wheel selection strategy is proposed
different vehicle states. If it is a two-wheel braking scheme, when based on ƛM and ƛγ . In Figure 7 [left front, right front wheel, rear left
the car is about to flick or understeer, the system will adopt the method and right rear wheel], when the value in the vector is set to 1,
of active braking the two wheels on the outer side of the vehicle’s corresponding tire brakes, when the value in the vector is set to 0, tire
rotation direction at the same time to adjust the body state. The two don’t brake, 0|1 said whether need the auxiliary brake wheel brake,

Frontiers in Robotics and AI 09 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

TABLE 1 Vehicle parameters.

Parameters Units Value

mass kg 1570

The moment of inertia of the body around the z-axis kg · m2 4192

Vehicle center of mass to front axle distance m 1.04

Vehicle center of mass to rear axle distance m 1.56

Vehicle front axle cornering stiffness Nm/rad −78329

Vehicle rear axle cornering stiffness Nm/rad −78329

MATLAB/Simulink, as shown in Figure 8. During the

operation of the collision avoidance algorithm, the vehicle is
driving at a speed of about 84 km/h. The obstacle car is located at
(40, −2) position, as shown in Figure 9B. At this time, the vehicle
FIGURE 11
Variation of lateral error with time. is about to have a frontal collision, and the collision avoidance
algorithm starts to run and the steering wheel is turned to the
right. As shown in Table 1, the simulated vehicle parameters are
selected from Carsim.
for example, if the right front wheel braking does not produce
enough yawing moment, we need to brake the right rear wheel to
generate enough yaw torque. If the vehicle is oversteering (ƛ<0), the Without the stability control
priority braking wheels are the front outer wheels of the vehicle.
When the vehicle understeers (ƛ>0), the priority braking wheel is the Without the stability control, the planning algorithm and the
rear inner wheel. Once the yaw moment provided by the front lateral control algorithm as shown in Figures 9, 10 perceive that
wheels is not enough, the remaining yaw moment can be generated there is a slow-moving car at a speed of 10 km/h 40 m ahead, and
by the selected auxiliary wheels. then start to perform lane change to avoid obstacles. After about
Finally, according to the selected braking wheels, the distributed 2 s, the ego vehicle avoids obstacles, and in about 6 s, the heading
four-wheel braking force can be obtained from Formula. 26. angle gradually stabilizes near 0, but the yaw angle speed
stabilizes slowly. At this time, the longitudinal velocity vx of
the ego vehicle remains basically unchanged.
Simulation experiments The tracking performance can be represented by the change
of lateral error (The distance between the vehicle and its
In order to verify the effectiveness of the algorithm in this projection to the planned path) with time, as shown in the
paper, a joint simulation model was built in Carsim and following Figure 11.

FIGURE 12
(A) Vehicle speed (B) Obstacle avoidance trajectory.

Frontiers in Robotics and AI 10 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

FIGURE 13
(A) Changes of yaw rate with time (B) Changes of heading angle with time.

FIGURE 14
Change of tyre braking force with time.
FIGURE 15
Trajectory planning.

With the stability control

The stability control intervenes within 2–4 s after the vehicle avoids stably according to the road centerline after the lane change. The change
the obstacle as shown in Figures 12, 13. At this time, the change of the of tyre braking force with time as shown in Figure 14.
yaw rate is quickly suppressed. At the same time, it can be seen that the
heading angle stabilizes rapidly to 1° in 2.5 s. But there is a certain
heading angle error. The reason why 2–4 s is chosen is due to the Trajectory planning
consideration of stability control on path tracking accuracy. Firstly,
before avoiding obstacles (that is, before 2 s, this time is calculated by On the premise of keeping other planning parameters
TTC), the stability control has a certain impact on obstacle avoidance, consistent, we changed Wpsmooth2 from 1000 to 20,000 for a
and may even lead to accidents, which should be avoided as much as total of 20 tests (indicated by different colored trajectories). It can
possible. Secondly, if the stability control intervention time is too long or be seen that our algorithm successfully avoided the obstacle car
the stability control intervenes when the yaw angle is large, the lateral ahead under the condition of considering the smooth trajectory
control algorithm cannot control the heading angle error to zero. This as shown in Figure 15, which proves the effectiveness of our
will cause the planning control algorithm to be unable to track the path algorithm.

Frontiers in Robotics and AI 11 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

6 Conclusion Author contributions

In this paper, the planning algorithm during steering and collision All authors listed have made a substantial, direct, and intellectual
avoidance is studied based on the information and data given by vehicle contribution to the work, and approved it for publication.
environmental information perception and vehicle state parameter
estimation The coordinated control of vehicle stability and safety is
studied. The paper has carried out the following work: 1. Considering Funding
the calculation efficiency and control requirements of the model, a
three-degree-of-freedom vehicle dynamics model and a linear tire Project supported by National Natural Science Foundation of
model are established. 2. The planning module is proposed by the China (Grant No. 52075388).
method of quadratic programming. This module will plan the driving
trajectory of the vehicle by comprehensively considering the constraints
and safety of vehicle execution. 3. Considering the vehicle trajectory
tracking performance and stability, LQR lateral control and adaptive
Conflict of interest
MPC control algorithms are proposed and the intervention time is
Author LZ was employed by Global Technology Co., Ltd.
proposed. 4. According to the results output by the MPC algorithm, the
The remaining author declares that the research was
four-wheel braking force is distributed to realize the vehicle collision
conducted in the absence of any commercial or financial
avoidance control under the comprehensive consideration of safety and
relationships that could be construed as a potential conflict of
stability. The results show that the planning algorithm in this paper can
interest.
give a safe and reliable collision-free motion trajectory, and the
proposed stability and safety coordination control algorithm can
track the collision avoidance trajectory with high precision and
stabilize the vehicle’s heading angle about 0.5s after avoiding obstacles. Publisher’s note
All claims expressed in this article are solely those of the authors
Data availability statement and do not necessarily represent those of their affiliated
organizations, or those of the publisher, the editors and the
The original contributions presented in the study are included in reviewers. Any product that may be evaluated in this article, or
the article/Supplementary Material, further inquiries can be directed claim that may be made by its manufacturer, is not guaranteed or
to the corresponding author. endorsed by the publisher.

References
Borrello, G., Raffone, E., Rei, C., and Fossanetti, M. (2020). “Trajectory Planning and Li, G., Yang, Y., Zhang, T., Qu, X., Cao, D., Cheng, B., et al. (2021). Risk
Vehicle Control at low speed for home zone manoeuvres,” in 21st IFAC World assessment based collision avoidance decision-making for autonomous vehicles in
Congress on Automatic Control - Meeting Societal Challenges), Berlin,Germany, multi-scenarios. Transp. Res. Part C-Emerging Technol. 122, 102820. doi:10.1016/j.
July12–17,2020, 15516–15523. trc.2020.102820
Chen, X., Wu, G., and Ren, M. (2020). Nonlinear model predictive control of Li, P., Pei, X., Chen, Z., Zhou, X., and Xu, J. (2022). Human-like motion planning of
autonomous vehicles considering dynamic stability constraints. SAE Int. J. Adv. autonomous vehicle based on probabilistic trajectory prediction. Appl. Soft Comput.
Curr. Pract. Mobil. 2 (5), 2974–2986. doi:10.4271/2020-01-1400 118, 108499. doi:10.1016/j.asoc.2022.108499
Cheng, Y., Hu, X., Chen, K., Yu, X., and Luo, Y. (2022). Online longitudinal trajectory Li, Z. H., Wang, P., Liu, H. H., Hu, Y. F., and Chen, H. (2021). Coordinated
planning for connected and autonomous vehicles in mixed traffic flow with deep longitudinal and lateral vehicle stability control based on the combined-slip tire
reinforcement learning approach. J. Intelligent Transp. Syst., 1–15. doi:10.1080/ model in the MPC framework. Mech. Syst. Signal Process. 161, 107947. doi:10.1016/
15472450.2022.2046472 j.ymssp.2021.107947
Coelingh, E., Eidehall, A., and Bengtsson, M. (2010). “Collision warning with full auto Liu, W., Xia, X., Xiong, L., Lu, Y., Gao, L., and Yu, Z. (2021). Automated vehicle
brake and pedestrian detection - a practical example of automatic emergency braking,” sideslip angle estimation considering signal measurement characteristic. IEEE Sensors J.
in 13th International IEEE Conference on Intelligent Transportation Systems), Funchal, 21 (19), 21675–21687. doi:10.1109/JSEN.2021.3059050
Portugal, 19-22 September 2010, 155–160.
Liu, W., Xiong, L., Leng, B., Meng, H., and Zhang, R. (2017). Vehicle stability criterion
Cui, Q., Ding, R., Wei, C., and Zhou, B. (2021). A hierarchical framework of research based on phase plane method. SAE Technical Paper 2017-01-1560. doi:10.4271/
emergency collision avoidance amid surrounding vehicles in highway driving. 2017-01-1560
Control Eng. Pract. 109, 104751. doi:10.1016/j.conengprac.2021.104751
Liu, W., Xiong, L., Xia, X., and Yu, Z. (2018). “Intelligent vehicle sideslip angle
Ge, L., Zhao, Y., Ma, F., and Guo, K. (2022). Towards longitudinal and lateral coupling estimation considering measurement signals delay,” in 2018 IEEE Intelligent Vehicles
control of autonomous vehicles using offset free MPC. Control Eng. Pract. 121, 105074. Symposium (IV), Changshu, China, 26-30 June 2018, 1584–1589.
doi:10.1016/j.conengprac.2022.105074
Rabhi, A. E. d., Hartani, K., Guettaf, Y., and Norediene, A. (2021). Robust
Gilbert, A., Petrovic, D., Pickering, J. E., and Warwick, K. (2021). Multi-attribute multimachine control for bisynchronous propulsion traction chain of an electric
decision making on mitigating a collision of an autonomous vehicle on motorways. vehicle. Sae Int. J. Veh. Dyn. Stab. Nvh 5 (2), 173–189. doi:10.4271/10-05-02-0012
Expert Syst. Appl. 171, 114581. doi:10.1016/j.eswa.2021.114581
Ren, M., and Wu, G. (2020). “Integrated strategy for automatic lane-changing
Han, I. C., Luan, B. C., and Hsieh, F. C. (2014). “Development of Autonomous decision and trajectory planning in dynamic scenario,” in ASME 2020 International
Emergency Braking control system based on road friction,” in 2014 IEEE International Design Engineering Technical Conferences and Computers and Information in
Conference on Automation Science and Engineering (CASE), New Taipei, Taiwan, 18- Engineering Conference, (V004T04A005).
22 August 2014, 933–937.
Schieben, A., Griesche, S., Hesse, T., Fricke, N., and Baumann, M. (2014).
Hang, J. Y., Yan, X. D., Li, X. M., and Duan, K. (2022). In-vehicle warnings for work Evaluation of three different interaction designs for an automatic steering
zone and related rear-end collisions: A driving simulator experiment. Accid. Analysis intervention. Transp. Res. Part F-Traffic Psychol. Behav. 27, 238–251. doi:10.
Prev. 174, 106768. doi:10.1016/j.aap.2022.106768 1016/j.trf.2014.06.002

Frontiers in Robotics and AI 12 frontiersin.org

Liu et al. 10.3389/frobt.2023.1120658

Tork, N., Amirkhani, A., and Shokouhi, S. B. (2021). An adaptive modified neural Xiong, L., Xia, X., Lu, Y., Liu, W., Gao, L., Song, S., et al. (2020). IMU-based automated
lateral-longitudinal control system for path following of autonomous vehicles. Eng. Sci. vehicle body sideslip angle and attitude estimation aided by GNSS using parallel
Technology-an Int. Journal-Jestech 24 (1), 126–137. doi:10.1016/j.jestch.2020.12.004 adaptive kalman filters. IEEE Trans. Veh. Technol. 69 (10), 10668–10680. doi:10.
1109/TVT.2020.2983738
Wu, G. (2021). Automotive theory. 3rd ed. Beijing: China Communications Press.
Yang, C., Chen, X., Lin, X., and Li, M. (2022). Coordinated trajectory planning for
Wu, G., Lyu, Z., and Wang, C. (2023). Predictive shift strategy of dual-clutch
lane-changing in the weaving areas of dedicated lanes for connected and automated
transmission for driving safety on the curve road combined with an electronic map.
vehicles. Transp. Res. Part C-Emerging Technol. 144, 103864. doi:10.1016/j.trc.2022.
SAE Int. J. Veh. Dyn. Stab. NVH 7 (1), 1–19. doi:10.4271/10-07-01-0001
103864
Xia, X., Xiong, L., Liu, W., and Yu, Z. (2018). “Automated vehicle attitude and lateral
Zha, Y., Quan, X., Ma, F., Liu, G., Zheng, X., and Yu, M. (2021). Stability control for a
velocity estimation using a 6-D IMU aided by vehicle dynamics,” in 2018 IEEE
four-wheel-independent-drive electric vehicle based on model predictive control. SAE
Intelligent Vehicles Symposium (IV), Changshu, China, 26-30 June 2018, 1563–1569.
Int. J. Veh. Dyn. Stab. NVH 5 (2), 191–204. doi:10.4271/10-05-02-0013
Xiao, W., Cassandras, C. G., and Belta, C. A. (2021). Bridging the gap between optimal
Zhang, R., Xiong, L., Yu, Z., and Liu, W. (2017). A nonlinear dynamic control design
trajectory planning and safety-critical control with applications to autonomous vehicles.
with conditional integrators applied to unmanned skid-steering vehicle. SAE Technical
Automatica 129, 109592. doi:10.1016/j.automatica.2021.109592
Paper 2017-01- 1585. doi:10.4271/2017-01-1585

Frontiers in Robotics and AI 13 frontiersin.org