An Efficient and Robust Improved A* Algorithm for Path Planning

Abstract

Path planning plays an essential role in mobile robot navigation, and the A* algorithm is one of the best-known path planning algorithms. However, the conventional A* algorithm and the subsequent improved algorithms still have some limitations in terms of robustness and efficiency. These limitations include slow algorithm efficiency, weak robustness, and collisions when robots are traversing. In this paper, we propose an improved A*-based algorithm called EBHSA* algorithm. The EBHSA* algorithm introduces the expansion distance, bidirectional search, heuristic function optimization and smoothing into path planning. The expansion distance extends a certain distance from obstacles to improve path robustness by avoiding collisions. Bidirectional search is a strategy that searches for a path from the start node and from the goal node at the same time. Heuristic function optimization designs a new heuristic function to replace the traditional heuristic function. Smoothing improves path robustness by reducing the number of right-angle turns. Moreover, we carry out simulation tests with the EBHSA* algorithm, and the test results show that the EBHSA* algorithm has excellent performance in terms of robustness and efficiency. In addition, we transplant the EBHSA* algorithm to a robot to verify its effectiveness in the real world.

An overview of path planning technologies for unmanned aerial vehicles

“…The purpose of A * algorithm is to find the shortest path and directs its search towards the shortest path states with a heuristic function. Thus, the efficacy of the A * algorithm is better than Dijkstra [22]. This algorithm is used in some cases in the dynamic environments.…”

An improved A star algorithm for wheeled robots path planning with jump points search and pruning method

“…Therefore, some researchers have proposed to improve the A* algorithm. For the purpose of stability and effectiveness, Wang et al [27] presented the improved method from the view of expansion distance, bidirectional search, heuristic function optimization, and smoothing. Song et al [28] considered the cost function with an additional parameter, with the result of fewer nodes and lower memory overhead.…”

A Smooth Jump Point Search Algorithm for Mobile Robots Path Planning Based on a Two-Dimensional Grid Model

“…Yang et al [7] extended the search method of the traditional A * algorithm from eight directions in eight neighborhoods to sixteen directions in sixteen neighborhoods at the expense of the search e ciency of the algorithm, which improved the smoothness and optimality of the path. Wang et al [8] used a two-way search strategy to improve the A * algorithm, and the simultaneous iterative search in both positive and negative directions improved the search e ciency, but it ignored the problem of large memory occupation in largescale map paths. Zafar et al [9] used the jump point search method to improve the node expansion method of the A * algorithm, which solved the memory problem of high overhead and low search e ciency in large-scale maps [8], but the smoothness of the paths is not satisfactory yet.…”

“…Wang et al [8] used a two-way search strategy to improve the A * algorithm, and the simultaneous iterative search in both positive and negative directions improved the search e ciency, but it ignored the problem of large memory occupation in largescale map paths. Zafar et al [9] used the jump point search method to improve the node expansion method of the A * algorithm, which solved the memory problem of high overhead and low search e ciency in large-scale maps [8], but the smoothness of the paths is not satisfactory yet. Tang et al [10] used third-degree B-spline to smooth the paths planned by the A * algorithm but ignored the constraint problem of the obstacle region, which is prone to cause the smoothed paths to traverse obstacles in a narrow area.…”

“…e above-improved algorithms [7][8][9][10][11][12][13] improve the search efficiency, path smoothing, and security of the traditional A * algorithm to some degree. Considering the above issues comprehensively, the main contributions of this paper are as follows: firstly, the obstacles modeled by the grid method are puffed, and the risk areas are defined, and then, our improved A * does not search the risk region during the search path so that we reduce the computation of the algorithm and ensure the safety of the planned path.…”