Autonomous Vehicle Trajectory Planning and Control Based on Traffic Motion Prediction

Abstract

The autonomous vehicle should be properly operated according to the road geometry and behavior of the surrounding vehicles. For this purpose, we propose a supervisor that predicts the motion of the surrounding vehicle using the artificial potential field method, determines the collision risk of the vehicle with the collision check algorithm and generates the velocity distribution map considering the velocity of the surrounding vehicles. In addition, the trajectory planner is designed to minimize the jerk for passenger comfort, considering the road shape and the velocity distribution determined by the supervisor. Finally, the nonlinear model predictive controller is formulated considering the limit of the actuator and collision avoidance.