A Novel Dynamic Obstacle Avoidance Strategy Based on Predictive Artificial Potential Field Algorithm

Ruo Chen,Yang Zhu,Bin Cao,Hongye Su

DOI: https://doi.org/10.23919/ccc63176.2024.10661442

2024-01-01

Abstract:Autonomous vehicles are playing an increasingly important role in production and life, but for some high-speed, malicious attempts to move obstacles, their safety performance can not meet the needs. Based on the artificial potential field (APF) method, an improved predictive artificial potential field (PAPF) method is proposed to avoid collision by predicting the collision point. Based on the intention of the obstacle to collide with the car, PAPF predicts the collision point and speed, and obtains the position of the virtual obstacle and its repulsive force. The virtual obstacle is the collision point predicted by the algorithm. At the same time, the attractive field is improved to prevent the car from falling into the local minimum point, and the effect of the attractive field is reduced to prevent the car from interfering with the obstacle avoidance behavior when the threat is too large. Simulation results show that PAPF method is feasible and has better security performance than traditional methods.