The paper aims to solve the problems of A* algorithm, such as a large number of nodes and time-consuming in search. The traditional A* algorithm and the bidirectional synchronization A* algorithm were improved, so that the algorithm can search the path from the starting point and the end point simultaneously, and improve the heuristic function. In order to verify the effectiveness of the improved algorithm, two-dimensional and three-dimensional simulation of warehouse AGV was carried out. The improved bidirectional synchronization A* algorithm reduced the time-consuming by 39.1%, 34.3%, 34.6% under different maps. The searching nodes can be reduced by 140 at most, and the path length was basically the same. The improved bidirectional synchronization A* algorithm is more efficient than the traditional A* algorithm in planning the optimal path, considering both the optimal path and the low travel complexity, improving the retrieval efficiency, and effectively solving the problems of more query nodes and more time-consuming.