Vehicle lateral control for driver assistance and automated driving.

摘要

In this dissertation, the issues on the lateral motion control of roadway vehicles are addressed for both the driver assistance and the automated driving systems.; The modeling of vehicle lateral motions is the fundamental aspect to be studied first. The two dominant motions for vehicle lateral control are the yaw and lateral motions. A two degree-of-freedom (DOF) model commonly used to describe these motions is called the bicycle model. However, experimental results for certain vehicles have shown some frequency characteristics that can not be explained by the bicycle model. A 3 DOF vehicle model, which incorporate the suspension roll dynamics, is developed and verified against the experimental data. The results attribute the discrepancy in the frequency characteristics to the vehicle suspensions, especially the roll dynamics. The fundamental coupling between the lateral, yaw, and roll dynamics is also addressed using the developed linear 3 DOF vehicle model.; A lateral guidance system is designed for driver assistance. A supplementary guidance display, which includes the road information, the current and future vehicle position, can provide the driver with the preview information. Two display laws for predicting the future position of the vehicle have been studied. The kinematic display law with simple formulation is applicable for low vehicle speeds, while the nonlinear gain-scheduled display law can be adopted to a wide range of vehicle speeds. With the assistance of the lateral guidance system, the driver may act as a simple proportional controller to safely steer the vehicle. The results of closed-loop simulation demonstrate the advantage of the guidance display. The human-in-the-loop steering simulator is used to further validate the effectiveness of the lateral guidance system for driver assistance.; There are two aspects being focused in the design of steering controllers for automated driving. In order to investigate the effect of suspension roll dynamics to the lateral motions, both the 3 DOF vehicle model and the bicycle model are utilized in the design of steering controllers. The μ-synthesis and H_∞ theory are applied to both vehicle models for a comparison study of robust steering controller design. The simulation results indicate the importance of the effect of roll dynamics to steering control, especially for vehicle with soft suspensions. On the other hand, the effectiveness of the look-ahead scheme for the steering controller design is also addressed in this dissertation. The look-ahead scheme can provide more degree of freedom in steering control comparing with the look-down sensing systems. The lateral dynamics is decoupled from the yaw and roll motions by the proposed look-ahead scheme. A velocity invariant steering controller is generated by using the decoupling control scheme while the vehicle is operated under different longitudinal speeds.