Vehicle Following: Control Design, Simulation and Experiments

Tuấn Minh Phạm (pmtuan@sti.vast.ac.vn)
School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore
November, 2006
Full text (external site)
 

Abstract

The thesis addresses on the development and implementation of vehicle-following controllers for a fleet of two vehicles, a leader vehicle and a follower vehicle. Critical issues are studied including theoretic development, simulation study and experimental implementation for the proposed controllers.

The thesis starts off with a comprehensive review of previous works on the vehicle-following control systems which are normally divided into two independent control designs: longitudinal control and lateral control; as well as the issue of fault detection and identification and fault-tolerance for such systems, with applications mainly to car-like and/or land wheeled vehicles.

Subsequently, this thesis presents its two main parts: the development and implementation of vehicle-following controllers for car-like vehicles and the development and implementation of a fault-tolerant vehicle-following controller for four-wheel-steering vehicles.

(+) Vehicle-following control for car-like vehicles:
The kinematics-based and dynamics-based vehicle-following controllers are developed based on the definition of a focus point in front or behind the follower vehicle. The focus point is a function of the vehicle position
as well as the steering angle. By controlling the follower vehicle in such a way that this focus point is able to track a reference point on the leader vehicle, the vehicle following is proven to be successful for both look-ahead forward tracking and look-behind backward tracking. The feedback control laws integrate both longitudinal control and lateral control as one to make it possible for the vehicle to achieve any maneuvers. The effects of system parameter selections are also investigated especially for two basic tracking maneuvers: straight and circular trajectories. The results reveal that the necessary
conditions to achieve tracking convergence are intuitive and similar to the human driving practices, for instance the driver looks ahead when following a car in front or he will not focus far away when engaging in a tight turn. The proposed controllers are then simulated on a pseudo-reality co-simulation platform.

On one hand, instead of modeling the vehicles by using differential equations, the vehicles and the working environment are modeled in ADAMS in such a way similar to their normal construction process in practice of assembling designed and selected components and materials. On the other hand, the control algorithms are programmed comfortably in SIMULINK, with the vehicle system model imported and represented by a functional block. The co-simulation platform provides a high-fidelity simulation environment that can replicate the actual vehicle system and produce almost precisely what the behaviours could be when the control algorithm is implemented. Based on this simulation platform, the vehicle-following controllers are extensively studied with different sets of control parameters for different tracking situations. The results validate the effectiveness of and offer interesting insights from the proposed controllers.
The controller is also implemented on the experimental vehicle-following system, Cycab vehicles. A sensing system is designed to detect and measure the relative distance and orientation between two vehicles. The sensing system utilizing a laser scanner and reflective tapes provides a fast updating rate of 30Hz, with the accuracy of 2cm for distance of up to 80m, and 0.5 degrees for the angle. Besides, by using three reflective tapes to define the tracked target, the reliability against noise and disturbances of the sensing system is significantly improved. The experiments are then carried out to illustrate the efficiency of the proposed controller.

(+) Fault-tolerant vehicle-following control for four-wheel-steering vehicles:
A fault-tolerant vehicle-following controller for four-wheel-steering vehicles is developed based on the one for car-like vehicles. The controller is able to carry on its vehicle-following task even if one of its driving systems or steering systems is at fault. The controller utilizes the additional steering actuator to switch among three steering modes, front-wheel-steering, rear-wheel-steering and four-wheel-steering, depending on the working condition of the two steering systems. The working principles of driving motors are exploited to gain the fault tolerance capacity for the driving control. The necessary conditions are constructed from the combination of conditions for each faulty situations. In order to implement the fault-tolerant controller, the low-level control system of the four-wheel-driving four-wheel-steering vehicle is redesigned using a decentralized control architecture. Both hardware and software are developed in such a way that when a fault occurs at any part of the system, it will be detected and isolated while the remaining parts of the vehicle continue to operate. Based on the physical structure of the vehicle, two separate but cooperative low-level controllers are built up to manage the front and
rear sets of driving and steering systems. Each controller by itself can manipulate the vehicle, if the other is not working. When both controllers are working, they communicate with each other to cooperate in controlling the vehicle. The vehicle-following maneuver is supervised by the proposed controller which integrates all the fault diagnostic information from the vehicle to adjust its parameters and control commands. Simulation and experimental results show the efficiency of the fault-tolerant vehicle-following system.