引用本文: 王立勇, 马少博, 王超, 丁炳超, 李伯雄, 王浩东, 苏清华. 基于目标导向和分层平滑优化JPS算法的移动机器人运动规划[J]. 机器人, 2023, 45(4): 439-450.
WANG Liyong, MA Shaobo, WANG Chao, DING Bingchao, LI Boxiong, WANG Haodong, SU Qinghua. Motion Planning of Mobile Robot Based on Goal-orientated Hierarchical Smoothing Optimization JPS Algorithm[J]. ROBOT, 2023, 45(4): 439-450.
 Citation: WANG Liyong, MA Shaobo, WANG Chao, DING Bingchao, LI Boxiong, WANG Haodong, SU Qinghua. Motion Planning of Mobile Robot Based on Goal-orientated Hierarchical Smoothing Optimization JPS Algorithm[J]. ROBOT, 2023, 45(4): 439-450.

## Motion Planning of Mobile Robot Based on Goal-orientated Hierarchical Smoothing Optimization JPS Algorithm

• 摘要: 针对JPS（跳点搜索）算法的搜索过程缺少方向引导，检索路径上存在较多冗余点、转折点，且路径不平滑的问题，提出了一种基于目标导向和分层平滑优化的跳点搜索（GHSO-JPS）算法，并通过多段多项式法和基于斥力势场的碰撞惩罚算法对运动轨迹进行优化。首先通过引力势场增强搜索方向的目标性，减少无关跳点的搜索。其次采用分层平滑优化策略消除路径中的冗余点和转折点，减少路径点数量和路径长度，提高路径平滑性。最后采用多段多项式方法进行轨迹优化，并通过斥力势场构建轨迹碰撞惩罚函数，以提高轨迹安全性。在实验室和校园区域对JPS算法、A*算法、RRT（快速扩展随机树）算法和本文算法进行实车实验对比和分析。结果表明，本文算法在4种算法中性能最佳。相比于传统JPS算法，在校园区域中本文算法的搜索时间缩短41.6%，跳点数量减少55.5%，路径长度减小3.22 m，并且路径点数量和总转折角度分别减小89.8%和81.8%。相比于A*算法和RRT算法，本文算法的规划时间分别缩短70.4%和93.7%，规划的最优路径的总转折角度分别减小了90.3%和97.1%。综上，该算法较传统运动规划方法具有更好的路径规划性能、更高的效率及更强的轨迹优化能力。

Abstract: Due to lack of direction guidance in the path search process of JPS (jump point search) algorithm, there are many redundant points, turning points and uneven paths on the planned path. Therefore, a GHSO-JPS (goal-oriented hierarchical smoothing optimization JPS) algorithm is proposed. Besides that, the multi-segment polynomial method and the collision penalty algorithm based on repulsive potential field are applied to trajectory optimization. By the attractive potential field, the search is focused on the goal direction, and irrelevant jump points in the search are reduced. Then, a hierarchical smoothing optimization strategy is adopted to eliminate redundant points and turning points in the path. Besides that, both the path point amount and the path length are reduced while improving the path smoothness. Finally, the multi-segment polynomial method is used to optimize the trajectory, and the penalty function against trajectory collision based on repulsive potential field is built to improve the trajectory safety. The real vehicle experiments are executed in the areas of laboratory and university, to analyze the proposed algorithm and compare with JPS, A*, RRT (rapidly-exploring random tree) algorithms. As the result, the proposed alogrithm shows the best performances among the 4 algorithms. Compared with the traditional JPS algorithm, the search time of the proposed algorithm is reduced by 41.6%, the amount of jump points is reduced by 55.5%, the path length is reduced by 3.22 m, and number of path points and total turning angle are reduced by 89.8% and 81.8% respectively, in the university area. Compared with A* and RRT algorithms, the planning time of the proposed algorithm is reduced by 70.4% and 93.7% respectively, and the total turning angle of the planned optimal path is reduced by 90.3% and 97.1% respectively. In conclusion, the proposed algorithm is of a better path planning performance, a higher efficiency and a better ability of trajectory optimization than the traditional motion planning methods.

/

• 分享
• 用微信扫码二维码

分享至好友和朋友圈