Predictive Obstacle-avoidance Control for Autonomous Vehicle

A predictive obstacle-avoidance control approach based on human driving behavior is presented in order to solve the problem of obstacle avoidance for autonomous vehicles. First,a safe-driving strategy based on human driving behaviors is proposed. Next,the desired lateral displacement and the desired longitudinal velocity are obtained by combining it with the search algorithm of the preview point. Finally,the kinematics formulae are adopted to calculate the actual value of the lateral displacement,and according to this,a linear model predictive controller is designed to control the vehicle's steering in order to track the desired trajectory. At the same time,a proportional integral derivative (PID) controller for longitudinal velocity control is designed to make the vehicle obtain the desired longitudinal velocity. To verify the validity of the proposed method,simulations under different road conditions are conducted. The simulation results show that the controlled vehicle can effectively avoid dynamic obstacles and vehicles during the running process.