This paper presents a new lateral control algorithm applicable to lane following, lane changing and obstacle avoidance for a vision-based automated vehicle that drives with detection of references including road edges and lane markers with the vision system. The steering angle is given by a coefficient of a cubic curve approximating a path in front of the vehicle, along which the vehicle is steered. The path is virtually generated in an onboard controller with the references and locations of obstacles. Experiments were conducted with a passenger car equipped with a stereo vision system, a personal computer-based controller, and a steering actuator. The automated vehicle drove at up to 30 km/h, taking various maneuvers of lane following, lane changing, and obstacle avoidance with the lateral control algorithm.
Abstract