Trajectory Optimization of Automatic Parking and Tracking Control Based on MPC

This paper aims to present an improved trajectory optimization method for auto-matic parking. An optimal control problem is frequently constructed to generate a sequence of discrete points for trajectory optimization of unmanned ground vehicles (UGVs). However, collision-avoidance constraints are always nonlinear with enormous amounts. To solve this problem, we propose a method for generating velocity adaptive safe corridors based on safe flight corridors (SFCs) to create linear collision-avoidance constraints. The maximum size and step of enlargement are determined by the velocity of each discrete point, which can im-prove solving speed of optimal control problem on the premise of avoiding collision. After getting the numerical solution of reference trajectory, we designed a tracking controller based on model predictive control (MPC) algorithm to conduct the simulation of trajectory tracking experiment. The tracking results prove that the trajectory is collision-free and kinematically feasible, and the MPC controller enables vehicle to track desired trajectory with high precision.