Robustness and performance issues in the lateral control of vehicles in automated highway systems

The aim of this dissertation is to advance the understanding of modeling and control of vehicles for automated lane following in Automated Highway Systems (AHS) framework. Several issues in the lateral control of vehicles in AHS are investigated. In particular, three areas are studied: (1) A dynamical model of the four wheeled road vehicle is derived with minimum assumptions as compared to the existing models. (2) Off-tracking, a phenomenon pronounced in vehicles with long wheel-base, is studied. Comparison is made between the kinematic and dynamic steady state behavior. It is concluded that the kinematic model of vehicles is a limiting case of the dynamic model. (3) Feedback controllers are designed for automatic lane following (lateral control) using steering as the control. One of the control objectives is to achieve acceptable performance in tracking of the lane centerline by a point on the vehicle. Furthermore, the relative yaw dynamics have to be damped such that there is no fish-tailing. The required performance is to be achieved in the presence of parametric model uncertainties. With this in view, the use of two representative robust control techniques, the Sliding Mode Control (SMC) and the {dollar}Hsb{lcub}infty{rcub}{dollar} robust control is studied. SMC approaches have been popular for vehicle control for some time. In this dissertation, the emphasis is on chatter elimination in SMC technique while retaining the asymptotic tracking of lane centerline. Towards this goal, a steering rate based Sliding Mode Control is designed. In this design, the integration of the discontinuously switching steering rate gives a chatterless steering angle control command. A comparison of this chatter elimination strategy with a better known strategy is done. It is shown that the fundamental difference between the two SMC strategies is in the location of an integrator in the feedback loop. Closed loop experiments show that the SMC is effective in tracking the lane centerline. However, at high speeds and with the location of lateral error sensor (used to measure the deviation from the road centerline) fixed, the yaw dynamics cannot be damped well. Two approaches are adopted to solve the problem of underdamped internal dynamics. In the first approach, the lateral error sensor is assumed fixed at the front end of the vehicle. A combination of input-output linearization and {dollar}Hsb{lcub}infty{rcub}{dollar} control techniques is used to incorporate robust-performance in the controller design. It is shown that this approach results in better damping of the yaw dynamics while maintaining acceptable lateral error. In the second approach, the location of lateral error measurement is allowed to vary with time. This measurement is computed using combination of multiple sensors, each fixed to the vehicle. It is shown that if the location of the lateral error measurement is a certain function of the longitudinal velocity, then the input-output linearization results in a critically damped closed loop behavior. Comparison between the controllers by simulations is presented to establish validity of the theoretical claims.