Abstract: Aiming at the problems that the traditional dynamic window approach（DWA） is difficult to select the optimal path and the generated path is not smooth in the dense obstacle area, an improved DWA obstacle avoidance algorithm for mobile robots is proposed. Firstly, the pre-trajectoriy of the robotthat does not intersect with the obstacle is selected based on the tangent vector of the differential manifold, then the obstacle number factor and the direction angle change factor are introduced to improve the evaluation function to improve the safety of the robot in the areawith dense obstacles. Finally, the improved evaluation function is used to evaluate the selected trajectories and then the speed corresponding to the optimal trajectory is determined. The comparison of several groups of simulation experiments shows that the improved DWA algorithm can plan a more reasonable and smoother running path in the area with dense obstacles, which not only ensures the safety of the robot, but also has better obstacle avoidance effect.

Key words: obstacle avoidance, dynamic window approach（DWA） algorithm, differential manifold, obstacle number factor, directional angle change factor

摘要： 针对传统动态窗口法（DWA）在稠密障碍物区域存在最优路径难以选取及生成路径不平滑等问题，提出了一种改进的DWA移动机器人避障算法。基于微分流形切向量选取与障碍物不相交的机器人预轨迹，引入障碍物数量因子与方向角变化因子来改进评价函数，提高机器人在障碍物密集区域运行的安全性，使用改进后的评价函数对选取的轨迹进行评价，进而确定最优轨迹对应的速度。通过多组仿真实验对比表明：改进的DWA算法在障碍物密集区域能规划出更合理、平滑的运行路径，在保证了机器人安全性的同时还具有更好的避障效果。

关键词: 避障, DWA算法, 微分流形, 障碍物数量因子, 方向角变化因子