Improved Safety-First A-Star Algorithm for Autonomous Vehicles

Abstract

The research focus of this paper is to improve the traditional A-star algorithm to meet the needs of autonomous vehicles for path safety and smoothness. First, the improved algorithm considers the factor of obstacle distance in the heuristic function. This allows the algorithm to strike a balance between path length and path security, as well as avoiding searching for redundant nodes that are too close to obstacles. Besides, the expansion mode of the 8-connection of the traditional A-star algorithm is improved to the 20-connection, so that the sharpness of turning at the corner can be greatly reduced. And because there is enough safety distance between the planned path and obstacles, the path smoothing can be performed to meet the vehicle dynamics. The improved safety-first A-star algorithm is compared with the traditional A-star algorithm in various scenes. The experimental results prove that the security of the improved safety-first A-star algorithm is greatly enhanced.