Improved Hybrid A* algorithm for Autonomous Driving Path Planning
Gongyu Shang, Gu Gong, Fan Jiang, Xintong Liu · 2024
Aiming at the problems of large search space and low computational efficiency of traditional $A *$ algorithm in the process of path search, this paper proposes an improved Hybrid A * algorithm fused with DL-IAPS algorithm for self-driving local path planning. By fusing the DL-IAPS algorithm with the Hybrid A * algorithm and modifying the weights of the distance approach algorithm to follow the warm start path, the optimal path planning for the self-driving car to avoid obstacles in the static obstacle scenarios is improved. The fastest path passing time in Scene 1 and Scene 2 are 31.24 s and 36.32 s, respectively, which is $29.50 \%$ higher than the path passing time before the improvement approximately. In addition, the optimized global path planning of the self-driving car has a smoother and more fluid route, and the speed of the driving process is more uniform than before, which improves the safety performance of the car driving.