In the process of obstacle avoidance, there are some problems such as low precision and poor safety in the path tracking of intelligent vehicles. To solve these problems, a model prediction controller with soft constraints is designed in this paper. The controller adopts the vehicle motion model, according to the radius of the circumcircle of the obstacle and the vehicle, and the anti-collision constraint condition is constructed to improve vehicle safety. The vehicle tracks its path using a model predictive control (MPC) algorithm. Sideslip angle and acceleration control variables were introduced as soft constraints to improve vehicle path tracking accuracy. The simulation was carried out by CarSim and MATLAB/Simulink platforms. The results show that the constrained MPC control algorithm is advanced in the accuracy and stability of trajectory tracking control under high-speed obstacle avoidance.