Optimization of the Pure Pursuit algorithm based on real-time error

Abstract

To address the issues of vehicle overshoot on high-curvature paths (with a curvature radius less than 15 m) and the poor tracking performance on continuous turning paths in the traditional Pure Pursuit (PP) algorithm, this paper proposes an optimized Pure Pursuit control algorithm based on real-time error (PP-PD). First, the product of the vehicle's real-time error absolute value and an error coefficient is introduced into the traditional look-ahead distance calculation formula. Second, the control output of the PP algorithm is combined with the proportional-derivative (PD) control result based on real-time error according to a weighted formula to form a closed-loop control structure. The validation experiments are conducted using co-simulation of CarSim and Matlab/Simulink. The average error is used as the tracking cost index, and the proportion of vehicle overshoot in time and space is used to quantify the overshoot situation. The simulation results show that compared with the traditional PP algorithm, the PP-PD algorithm significantly improves tracking performance on continuous turning and high-curvature paths: the tracking cost is reduced by 43 %, and the overshoot is reduced by 51 %. This fully demonstrates that the PP-PD algorithm can greatly reduce vehicle overshoot and significantly improve tracking accuracy on high-curvature sections, providing a solid foundation for the smooth operation of the vehicle.