Research on Trajectory Planning and Tracking Strategy of Lane-changing and Overtaking based on PI-MPC Dual Controllers 2021-01-1262

Aiming at the problem of poor robustness after the combination of lateral kinematics control and lateral dynamics control when an autonomous vehicle decelerates and changes lanes to overtake at a certain distance. This paper proposes a trajectory determination and tracking control method based on a PI-MPC dual algorithm controller. To describe the longitudinal deceleration that satisfies the lateral acceleration limit during a certain distance of lane change, firstly, a fifth-order polynomial and a uniform deceleration motion formula are established to express the lateral and longitudinal displacements, and a model prediction controller (MPC) is used to output the front wheel rotation angle. Through the dynamic formula and the speed proportional-integral (PI) controller to control and adjust the brake pressure. Based on simulation to optimize the best lane change completion time coefficient at different longitudinal lane change speeds, the relationship between the vehicle collision avoidance stable lane change time and the real-time vehicle speed and deceleration is obtained, then it is optimized by neural network algorithm, to avoid the vehicle collision avoidance and deceleration change unstable performance such as rollover occurred during the road. Finally, the simulation verification of the deceleration and lane changing to overtake conditions at a certain initial vehicle speed shows that the maximum lateral acceleration is 3.03m/s², and the error from the maximum allowable acceleration is 1%. The maximum error of the yaw angle is 0.8°, and the maximum lateral acceleration is 3.22m/s² and 3.16m/s² respectively, which does not exceed the allowable acceleration of 4m/s², which satisfies the lateral stability of the vehicle. Therefore, in the study of trajectory planning and tracking control of autonomous vehicles, the controller can improve the control robustness of decelerating and changing lanes.