A self-adaptive safe A* algorithm for AGV in large-scale storage environment

Abstract

This paper presents a safe A* algorithm for the path planning of automated guided vehicles (AGVs) operating in storage environments. Firstly, to overcome the problems of great collision risk and low search efficiency in the path produced by traditional A* algorithm, a new evaluation function is designed by introducing repulsive term and assigning dynamic adjustment weights to heuristic items. Secondly, a Floyd deletion algorithm based on the safe distance is proposed to remove redundant path points for reducing the path length. Moreover, the algorithm replaces the broken line segments at the turns with a cubic B-spline to ensure the smoothness of turning points. The simulation applied to different scenarios and different specifications showed that, compared with other three typical path planning algorithms, the path planned by the proposed safe A* algorithm always keeps a safe distance from the obstacle and the path length is reduced by 1.95\(\%\), while the planning time is reduced by 25.03\(\%\) and the number of turning point is reduced by 78.07\(\%\) on average.