A collision avoidance strategy based on enhanced JPS and adaptive DWA

To address the limitations of the traditional dynamic window algorithm (DWA) in intelligent vehicle local path planning, such as insufficient foresight and difficulties in path selection, this paper proposes an improved DWA path planning algorithm for intelligent vehicles. By incorporating a bidirectional JPS algorithm, key nodes along the global path are generated as temporary target points for the local path planning of DWA, achieving global path optimization and real-time obstacle avoidance. The initial heading angle and velocity window of the traditional DWA algorithm are dynamically adjusted based on environmental conditions. The velocity evaluation function is enhanced, and a distance evaluation term from the local path endpoint to the target point is added to optimize the vehicle’s motion trajectory. Adaptive weight coefficients for DWA are implemented using ANFIS to improve path planning efficiency. Simulation experiments conducted in MATLAB across various environments validate the improved algorithm’s effectiveness and superiority. On a 20 × 20 map, the IDWA algorithm reduced path length by 14.4%, execution time by 11.81%, and improved path smoothness by 75.33% compared to the traditional DWA algorithm. On a 30 × 30 map, the IDWA algorithm reduced path length by 9.53%, execution time by 9.00%, and improved path smoothness by 73.84% compared to the traditional algorithm. On a 40 × 40 map, the IDWA algorithm achieved a 9.42% reduction in path length, a 12.88% decrease in runtime, and a 60.25% improvement in path smoothness compared to the traditional algorithm. While the traditional DWA algorithm may fail to plan in dynamic environments, the IDWA algorithm successfully reached the target in every simulation run. Furthermore, experiments conducted in real-world environments validated the effectiveness of the proposed algorithm.