关键词: 路径规划, A*算法, 邻域拓展, 平滑处理

Abstract: The traditional A* algorithm only has 8 searching-direction, which leads to the long and uneven path searching result. In order to solve this problem, the traditional A* algorithm is improved in two aspects in this paper. Firstly, the traditional 8-neighborhood A* algorithm is extended to 24-neighborhood, and then the guiding vectors are used to reduce the invalid neighborhood search and improve the efficiency. The vector is formed by the nodes, which have the minimum value in evaluation function of [f(n)]. Secondly, path smoothing algorithm is used to eliminate the redundant nodes of the path. Under the grid map environment and real scenes map, the improved A* algorithm is applicable. Compared with the traditional A* algorithm, the 24-neighborhood A* algorithm, the Dijkstra algorithm and the rapidly-exploring random trees（RRT） algorithm, according to the results of simulation experiments in Matlab, the improved A* algorithm has great advantages in performance items, such as the success rate, path length and searching time, and also, it performs well in the real scene map.

Key words: path planning, A* algorithm, neighborhood extension, smoothing