摘要: Trajectory tracking accuracy is of great significance for the control and safety management of autonomous vehicles. The curve with large curvature is often the bottleneck of vehicle control and the obstacle of safe driving. Aims a... 展开 Trajectory tracking accuracy is of great significance for the control and safety management of autonomous vehicles. The curve with large curvature is often the bottleneck of vehicle control and the obstacle of safe driving. Aims at the problem of partial visual field loss and low control accuracy of formula car in the process of driving on the curve with large curvature. This paper presents a trajectory planning and tracking framework, using the improved Delaunay triangle subdivision algorithm in cone bucket guide track path planning and real-time planning, combined with the neural network PID feedback the MPC control algorithm for global path tracking, experimental results show that the success rate of the improved path planning algorithm is increased by 55.8% on large curvature curves; The tracking control algorithm is superior to the traditional control algorithm in terms of tracking accuracy, which can reduce the steady-state error caused by the simplicity of the superior model and make the steady-state error close to zero faster. Compared with the traditional MPC, the overall average tracking accuracy is improved by 35.78% and the maximum lateral error is reduced by 52.99%. 收起