Research on Path Tracking Control of Unmanned Vehicles based on

Vision

Kaixin Yang

1, a

, Yunlong Bai

1, b

, Jianguo You

1, c

, Haibo Dong

1, d

and Jinwei Guo

1, e

Automotive Data Center, China Automotive Technology & Research Center Co. Ltd, Tianjin 300000, China

guojinwei@catarc.ac.cn

Keywords: Unmanned driving; Path tracking; Visual navigation.

Abstract: With the in-depth study of unmanned vehicle technology, highly efficient, stable and low-cost trajectory

tracking control system has become the key technology for intelligent and practical unmanned vehicles. The

technical level of automobiles and their popularity are important indicators to measure the material living

standard and modernization level of a country or region. Path tracking and motion control play a very

important role in intelligent unmanned driving technology. At the same time, accurate tracking of the expected

feasible path and stable motion control are the basis for realizing intelligent unmanned driving. The task of

path tracking is to control the vehicle to drive along the planned path, and at the same time to ensure the safety

and handling stability of the vehicle. The precise control of driverless vehicle is obviously the key technology

in the research and development of automatic driving. Therefore, based on the vision navigation system, this

paper studies the target path tracking control of driverless vehicle.

1 INTRODUCTION

The technical level of automobiles and their

popularity are important indicators to measure the

material living standard and modernization level of a

country or region. With the progress of science and

technology and the rapid development of economy

and society, the living standard of human beings is

improving day by day. Cars enter the homes of

ordinary people rapidly, which greatly facilitates

people's daily travel (Zhang Jiaqi, 2017). With the

quickening pace of modern life, frequent traffic

accidents caused by rapidly increasing car ownership

and increasingly serious driving fatigue have severely

restricted people's good life (Li Wei, Tang Zheng,

Wang Hongmin, 2019). Due to the limitation of

human's own perception ability and congenital

reaction delay, the incomplete and delayed grasp of

external information is easy to lead to wrong

judgment and wrong operation. In order to improve

the safety and stability of vehicles, make efficient use

of vehicles and roads, and reduce road congestion and

environmental pollution, the slogan of Intelligent

Transportation System (ITS) arises spontaneously.

Unmanned vehicles play an important role in ITS

system. Path tracking is one of the main research

contents of driverless cars. Its content is to design a

controller to ensure the car to follow a preset

trajectory. Under a certain control architecture, the

path tracking of unmanned vehicles can accurately

track real-time road information, and can follow the

planned virtual path without real-time road

information (Li Yongdan, et.al, 2019).

In the process of modernization, the automobile

industry has developed rapidly. Electronic

technology, computer technology and control

technology have injected fresh vitality and power into

the automobile industry. Cars bring great

convenience to people's lives and improve the quality

of life. As an intelligent comprehensive body, the

automobile has made indelible contributions to the

progress of human civilization and social

development due to its unique superiority. The

unmanned vehicle is a typical four-wheeled mobile

robot, involving many interdisciplinary knowledge,

and is a product of the highly integrated and

development of contemporary computer science,

pattern recognition, and control technology (Wang

Zizheng, Cheng Li, 2016). Unmanned driving

technology mainly includes several key parts:

environment perception, path planning, path tracking

and bottom-level control. The task of path tracking is

to control the vehicle to travel along the planned path,

Yang, K., Bai, Y., You, J., Dong, H. and Guo, J.

Research on Path Tracking Control of Unmanned Vehicles based on Vision.

DOI: 10.5220/0010061500640069

In Proceedings of the International Symposium on Frontiers of Intelligent Transport System (FITS 2020), pages 64-69

ISBN: 978-989-758-465-7

while ensuring the safety and stable operation of the

vehicle. Sex (Lan Jing, 2019). Unmanned driving

obtains the vehicle's own position and posture and

surrounding environment information through

various modern sensors. It has a high degree of self-

adaptive and self-planning capabilities, and can

obtain accurate information about its own posture and

surrounding environment through the sensor. The

trajectory tracking control system for driverless cars

based on visual navigation is of great urgency and

necessity (Zhao Chunchang, et.al, 2019). The precise

control of driverless cars is obviously the key core

technology in the research and development of

autonomous driving. Therefore, based on the visual

navigation system, this paper has carried out related

research on the target path tracking control of

driverless cars.

2 UNMANNED VEHICLE

ARCHITECTURE

With the progress of communication technology,

microelectronics technology and the development of

intelligent construction, more and more equipment

such as household appliances, electricity meters,

industrial terminals and so on have the demand of

networking, and a large amount of data need to be

transmitted. On many occasions, wired connection

cannot meet people's needs. Wireless access makes

up for the shortage of wired connection with its

advantages of convenience, quickness and low cost.

With the development of electronics, computer and

information science and technology, the development

of communication system is also very rapid. From

wired to wireless, from voice to data, images, from

local to wide area, there are many communication

networks and communication methods used in

unmanned vehicle positioning system. The sensors of

the unmanned vehicle are mainly used to sense the

environmental information around the intelligent

unmanned vehicle and the body state information of

the intelligent unmanned vehicle (Zhang Haiming,

et.al, 2019). In addition to autonomous navigation,

unmanned vehicle positioning and navigation

systems all need the support of communication

networks to facilitate the transmission of various data

and information. The whole system is designed with

the idea of modularization, and the main functions are

divided into independent modules, thus reducing the

complexity of the system and making the functions of

each module of the system clear at a glance.

The on-board vision system of unmanned vehicles

can effectively identify lane lines, traffic signs and

traffic lights, and can also detect obstacles in driving

lanes and adjacent lanes. As shown in Figure 1, the

system architecture takes a hierarchical form. With

high cohesion and low coupling, the output of the

previous layer is the input of the following layer. By

decomposing the work from top to bottom, the

accuracy of the design system is improved.

Figure 1: Hierarchical system architecture.

The available error correction codes of the system

can be shared by all users of the system and have the

function of automatically selecting error correction

codes. It is a multi-purpose and high-performance

wireless dispatching communication system that

shares resources, costs and error correction coding

equipment and services. The actions generated by the

hierarchical system are not the direct feedback

processing results of the data collected by the sensors,

but are processed through the stages of perception,

modeling, planning and control. The effective

connection between transportation and the whole

social and economic system is realized through

information technology. The integration of

transportation information into the information flow

of the supply chain will promote transportation to

truly become an organic part of the supply chain.

Establishing a global environment is not only based

on the user's understanding of the known objects in

the environment and the analysis of their

Research on Path Tracking Control of Unmanned Vehicles based on Vision

relationships, but also based on the independent

construction of sensor data. The global environment

model is universal and suitable for planning many

tasks. Special short-range communication system

mainly uses special short-range communication

technology to identify relevant information of

vehicles through signal transmitting and receiving

devices of roadside units, and automatically carries

out intelligent management such as identity

identification, real-time monitoring, dynamic

guidance and the like on vehicles to complete

dynamic collection of relevant information of

vehicles. In order to carry out communication in an

orderly way and to obtain the maximum successful

communication, some mechanism is needed to

arbitrate and decide who has the right to use error

correction codes.

3 PATH TRACKING CONTROL

OF UNMANNED VEHICLES

3.1 Motion Model of Unmanned

Vehicle

The main sensor of the lane line detection module is

a vehicle-mounted monocular camera, whose main

function is to capture the real-time lane line of the

vehicle while it is running. Firstly, the area where the

lane line is located in the camera image coordinate

system is determined, then the lane line position

information is obtained by feature extraction, and

then the coordinate is transformed to the vehicle body

coordinate system. The vehicle data transmission

system, which is composed of data transmission

based on conventional communication network, has

an effect range related to the band of error correction

coding, antenna height of central station and

transmission power. Two methods of wireless

transmission are simulated by adding convolutional

codes respectively, and the number of error code

frames in the simulation is recorded. The simulation

conditions are shown in Table 1.

Table 1: Simulation conditions.

Condition category Numerical value

Bit rate 850bps

Maximum Doppler shift 9.25Hz

Speed of vehicle 50km/h

Carrier frequency 445MHz

Signal to noise ratio 20dB

Error correction coding

parameters

Modulation system 2FSK

Suppose the longitudinal displacement of the car

at time k is x

, the lateral displacement is y

, and the

yaw angle is j

. Assuming that the acceleration of the

car is constant in a short time, as shown in Figure 1,

the displacement of each point is x

k-3

, ..., x

k + 1

, the

instantaneous speed is v

k-3

, ..., v

k + 1

, then:













































1111

3131

xxvv

kkkk

(1)

From v

-v

k-1

= v

k-1

-v

k-2

, we get:

231









kkkk

xxxx

(2)

If the longitudinal displacement of the vehicle at

time k-3,..., k is known, the predicted value of the

longitudinal displacement at time k+1 can be

obtained:

231









kkkk

xxxx



(3)

As a bearer network, ITS system itself adopts a

network structure and assigns independent addresses

such as or addresses to users. When the error

correction code is 1, the blocking rate is as shown in

Figure 2. The blocking rate when the error correction

code is 4 is shown in Figure 3.

FITS 2020 - International Symposium on Frontiers of Intelligent Transport System

Figure 2: Blocking rate when the error correction code is 1.

Figure 3: Blocking rate when the error correction code is 4.

In the automatic driving mode, since the steering

wheel is not controlled by the driver, there is no need

to detect the steering wheel input torque (Su Fandi,

2016). However, in order to realize the closed-loop

control of the automatic steering system, the steering

angle information of the steering wheel needs to be

detected in real time as the feedback quantity. The

traditional organization form of the transportation

industry, the traditional transportation management

mode, the traditional transportation concept and the

traditional transportation technology will all undergo

changes due to the construction of the intelligent

transportation system. Considering that the external

environment of the intelligent urban traffic

information system in practical application may be in

the downtown area, all kinds of electromagnetic

interference around it are more serious, and the same

frequency interference is even more serious.

Therefore, it is particularly important to improve the

communication reliability of the whole system. The

planning layer acquires the data of the decision layer,

generates a planning path from the initial point of the

vehicle to the punctuation point, and periodically

changes and updates the path of the vehicle according

to the environmental information.

3.2 Path Tracking Controller

If an unmanned vehicle encounters an emergency

during driving, it can execute an emergency stop

command to ensure the safety of the vehicle. The

remote control receiving module on board the

unmanned vehicle receives the signal and transmits it

to the microprocessor, thus controlling each

execution module on the bottom floor to realize

emergency stop. In order to reduce the complexity of

the intelligent urban traffic information system, the

centralized control error correction coding access

method is generally selected (Fan Rongwei. Yanfang,

2016). Because the communication in the system has

the characteristics of short time and small amount of

data, it is not suitable to adopt the reservation mode,

and usually the circular and competitive modes are

selected. If the acceleration of the vehicle is too large

or there is a sudden change, the body will have a large

impact, making passengers in the vehicle

uncomfortable. The control system should limit the

Research on Path Tracking Control of Unmanned Vehicles based on Vision

Figure 4: Packet blocking rate when voice call is 0.08.

Figure 5: Packet blocking rate when voice call is 0.12.

maximum acceleration range to make the vehicle

more stable. From the perspective of the whole

system, intelligent transportation is the embodiment

of many technologies. From the perspective of the

system as a whole, intelligent transportation is the

embodiment of many technologies. The receiving end

checks the relationship between the information

symbol and the supervised symbol according to the

established rules. Once an error occurs during

transmission, the relationship between the

information symbol and the supervised symbol is

destroyed, so that errors can be discovered and even

corrected.

Change the arrival rate of packet calls and voice

calls, and observe the blocking rate of packet calls.

Figure 4 shows the packet blocking rate when the

voice call arrival rate is 0.08. Figure 5 is the packet

blocking rate when the voice call arrival rate is 0.12.

The receiving end checks the relationship between

the information symbol and the supervision symbol

according to the established rules. Once errors occur

in the transmission process, the relationship between

the information symbol and the supervision symbol is

destroyed, thus errors can be found and even

corrected (Yu Jiawei, Luo Feng, 2016). The dynamic

model of a vehicle usually studies the mechanical

characteristics of the vehicle, such as tire mechanics,

vehicle driving mechanics, vehicle aerodynamics,

etc., and strives to establish the expression of the

relationship between the acceleration and the stress of

the vehicle, so as to find out the control rate of the

vehicle.

FITS 2020 - International Symposium on Frontiers of Intelligent Transport System

4 CONCLUSIONS

With the increasing annual production of

automobiles, the following major problems are traffic

congestion, energy consumption and automobile

safety. Since driverless vehicles completely exclude

the influence of human factors, they can greatly solve

the problem of traffic safety. In this paper, the vision-

based trajectory tracking control system in unmanned

driving technology is mainly studied, including

overall scheme formulation, hardware construction,

algorithm research and controller design. In the

design of electronic control system, some electric

equipments are very sensitive to low voltage due to

more electric equipments and serious power

consumption when the intelligent unmanned vehicle

runs autonomously in actual road environment.

Because the communication in the system has the

characteristics of short time and small amount of data,

it is not suitable to use the reservation mode, usually

choose the circular and competitive mode. If the

driverless vehicle encounters an emergency in the

process of driving, it can execute the emergency stop

order to ensure the safety of the vehicle. A stable and

reliable path tracking system is the necessary

condition for the industrialization of intelligent

vehicle. Subsequently, the control algorithm needs to

be transplanted to other embedded systems and fused

with multi-sensor data to realize the real-time control

of the vehicle, so as to verify the stability of the

control algorithm.

REFERENCES

Fan Rongwei. Yanfang Line fully automatic driverless

vehicle braking system [J]. Railway rolling stock, 2016

(5): 107-109.

Lan Jing. Development status and key technical analysis of

driverless cars [J]. Internal Combustion Engine and

Accessories, 2019 (15): 209-210.

Li Wei, Tang Zheng, Wang Hongmin. Research on path

tracking based on quasi-sliding mode control [J].

Journal of Chongqing Jiaotong University (Natural

Science Edition), 2019, 38 (10): 127-132.

Li Yongdan, Ma Tianli, Chen Chaobo, et al. Overview of

unmanned vehicle path planning algorithms [J].

Foreign Electronic Measurement Technology, 2019

(6): 72-79.

Su Fandi. Thoughts on fully automatic driverless cars [J].

Time Auto, 2016 (10): 49-50.

Wang Zizheng, Cheng Li. Introduction of driverless cars

[J]. Times Cars, 2016, 27 (8): 82-85.

Yu Jiawei, Luo Feng. Research on trajectory tracking

algorithm based on high-precision positioning system

[J]. Mechatronics, 2016 (10): 3-7.

Zhang Haiming, Gong Jianwei, Chen Jiansong, et al.

Unmanned vehicle car following method in

unstructured environment [J]. Journal of Beijing

Institute of Technology, 2019 (11): 1126-1132.

Zhang Jiaqi. A compound path planning algorithm based on

moving sub-goals [J]. China Journal of Highway and

Transport, 2017, 30 (11): 138-146.

Zhao Chunchang, Wen Chengwei, Zheng Hui. Linear

optimization based on driving safety and unmanned

driving [J]. Gansu Science and Technology, 2019 (11):

13-16.

Research on Path Tracking Control of Unmanned Vehicles based on Vision