Abstract: Aiming at the obstacle avoidance and path tracking control of intelligent vehicle, a model predictive controller (MPC) is designed. A layered structure is used in the controller. The high-level layer carries out the local path planning based on non-linear vehicle point mass model and the low-level layer realizes the path tracking based on monorail vehicle dynamics model. Through the joint simulation of Carsim and Matlab software, the controller parameters are optimized on-line by using Particle Swarm Optimization, and the performance of the controller is tested at different speeds. Simulation results show that the controller has good robustness, real-time performance and small tracking error. In the case of single and multiple obstacles, the hierarchical controller could avoid obstacles smoothly and re-track the original path in high accuracy.

Key words: intelligent vehicle, MPC, parameter optimization, path tracking, obstacle avoidance