MPC-Based Path Tracking Controller Design for Autonomous
Ground Vehicles
Chen Shen
1,2
,Hongyan Guo
1,2
, Feng Liu
2
, Hong Chen
1,2
1. State Key Laboratory of Automotive Simulation and Control, Jilin University (Campus NanLing), Changchun 130025, PR China
E-mail: chenh@jlu.edu.cn
2. College of Communication Engineering, Jilin University (Campus NanLing), Changchun 130025, PR China
Abstract: It is an important and essential aspect for autonomous ground vehicles to follow the desired path. In this manuscript,
a path tracking controller using model predictive control (MPC) method is proposed. In order to describe the vehicle motion and
its dynamics, kinematics and dynamics integrated model is first presented. Then, the control strategy is proposed and MPC is
employed to design the path tracking controller. In the following, the stability analysis is carried out and it shows that the system
is proven to be asymptotically stable and theoretically it has no static error. In order to validate the effectiveness of the proposed
MPC-based path tracking controller, veDYNA-Simulink joint simulations under different velocities and road friction coefficient
are carried out and the results illustrate that the path tracking algorithm obtains good tracking performance.
Key Words: Path tracking, MPC, Stability
1 Introduction
With the increasingly serious problem of traffic conges-
tion and road safety, autonomous vehicle has become an in-
creasingly important topic. An essential part of vehicle au-
tonomy consists of path planning and tracking control that
enables it to safely maneuver under different conditions[1].
Based on the road information, path planning aims to ob-
tain an anti-collision path according to environmental fac-
tors. Then, in path tracking, the key is to design the velocity
and steering controller to follow the established reference
path through velocity and steering control. Due to the strong
coupling, steering and velocity control are discussed in re-
spective, which have been described in great detail by many
authors.
The early discussed methods directly connect the vehi-
cle steering to the lateral position error. Then, W. Junmin
presented a steering control system in which vehicle yaw
rate is actively controlled to achieve trajectory tracking. [2].
Moreover, a nested PID steering control is proposed which
consists of two algorithms including inter-loop and outer
loop [3]. The outer loop is used to calculate the ideal yaw
rate depending on lateral position and the velocity and later-
al acceleration of autonomous vehicle are taken into consid-
eration in the inter-loop.
All the methods mentioned above simply depend on the
accuracy of the current environmental parameters and is
tough to take dynamics constraints into consideration. With
the rapid development of the computing power of hardware,
researchers proposed the model predictive control which has
significant advantages in dealing with the constraints in path
tracking. The real-time path planning and tracking method
based on predictive error feedback control as well as fuzzy
logic control are proposed, which is adaptive to the uncer-
tainty of road condition[4]. A receding horizon optimization
framework is proposed which used the vector of thrust forces
and moments as control input and by the nonlinear model
corresponding author
This work was supported by National Nature Science Foundation of
China (Grant No. 61520106008, 61403158, 61603148) and Project of the
Education Department of Jilin Province (Grant No. 2016-429).
predictive method, the depth and orientation of the AUV can
also be taken into consideration.[5]. In addition, in order
to obtain better control accuracy and flexibility, the robust
model predictive control has been proposed to eliminate the
position error and heading error. Thaker proposed a multi-
switch prediction model control scheme based on dynamic
compensation for trailer according to different velocity and
variable slip rate [6].
However, almost all of these control algorithms are ap-
plied in practice without in-depth theoretical analysis. In
conclusion, this paper presents an MPC based tracking con-
troller and theoretical analysis about stability and static error.
Assuming that the longitudinal velocity is constant, an MPC
based rolling horizon scheme is first proposed focusing on
lateral control, which establishes the kinematic and dynamic
state-space models of the vehicle. The front steering angle is
used as the input, and the lateral displacement is established
as the output. In the formulation of the control strategy, the
primary task is to reduce the error between an ideal path and
real path for path tracking. Then, the fuel consumption is
considered transformed to shortest path, besides, the actua-
tor’s action should be decreased within a diminutive range.
Moreover, three weight coefficients proposed aims to bal-
ance the importance of the three parts respectively. The sta-
bility analysis proves that the system is asymptotically stable
and the model error is zero without interference. To prove
the effectiveness of the controller, veDYNA-Simulink Joint
simulations are conducted at last.
The remainder of this paper is organized as follows. In
section II, the processes of establishing kinematic and dy-
namic model are described. Section III presents the MPC
based path-tracking controller for the autonomous vehicle.
In order to verify the performance of the controller, simula-
tions are carried out in section IV. The conclusions are given
in section V.
2 Vehicle Model
The proposed control algorithm for path tracking relies
on the MPC, thus the 2-DOF bicycle linear vehicle models
are required to generate the desired four parameters includ-
Proceedings of the 36th Chinese Control Conference
Jul
26-28, 2017, Dalian, China
9584