Abstract: Aiming at the disadvantages of traditional Rapidly Random Tree(RRT)algorithm in planning path,such as large randomness,low search efficiency,and the planned path is not conducive to robot movement,the improvement was made from three directions.For the problem of large randomness in the random tree expansion,the traditional expansion direction was added into the improved artificial potential field method to make the random tree grow in favor of the target point.Key points were extracted from the path planned of the improved RRT algorithm,and the path was optimized.The optimized path was segmented according to the key points using the improved evaluation function dynamic window method.Experiments showed that the optimized and improved RRT algorithm was better than the traditional A* algorithm and the traditional RRT algorithm in terms of path length,path planning time and inflection point,etc.The path planned by the fusion algorithm in complex environment could well avoid obstacles,and the path was smoother and shorter.

Key words: improve artificial potential field, optimize and improve rapidly random tree algorithm, improve dynamic window approach, fusion algorithm, obstacle avoidance

摘要： 针对传统快速搜索随机数(RRT)算法在规划路径中随机性较大,搜索效率较低且规划的路径不利于机器人移动等缺点,从3个方向进行改进。首先,对于随机树扩展时随机性较大的问题,将传统的扩展方向加入改进人工势场法约束,使得随机树偏向目标点生长;其次,将改进RRT算法规划的路径进行关键点提取,并优化路径;最后,将优化后的路径按照关键点分段使用改进评价函数的动态窗口法。实验表明,优化改进RRT算法相较于传统A*算法、传统RRT算法在路径长度、路径规划时间以及拐点等方面效果都更好,融合算法在复杂环境中规划出的路径能够很好地避开障碍物,路径更加平滑且更短。

关键词: 改进人工势场法, 优化改进RRT算法, 改进动态窗口法, 融合算法, 避障