【摘要】 为更好地解决移动机器人路径规划问题,面对使用改进Q-learning算法规划的路径存在拐角数量多、累计转弯角度大等不足,提出一种基于天牛须搜索算法的Q-learning路径优化方法(IBQL)。在传统天牛须搜索算法的基础上,引入运动步长递减策略提升天牛的搜索效率;添加障碍物碰撞检测并改进适应度函数,保证路径的安全性;按照运动最优化原则更新路径点,以降低位置更新的盲目性。仿真结果表明,IBQL算法的路径规划效果优异,使用改进天牛须搜索算法优化路径对算法整体的实时性影响很小,且优化后的路径具有拐点少、平滑性好等优点。更多还原

【Abstract】 In order to improve the path planning results of mobile robots and solve the problems of many corners and large cumulative turning angles in the path planned by the improved Q-learning algorithm,a Q-learning path optimization method( IBQL) based on beetle antennae search algorithm is proposed. The IBQL algorithm mainly improves the beetle antennae search algorithm in three aspects. First,the step-decreasing strategy is introduced to improve the efficiency of beetle antennae search. Second,it adds obstacle collision detection and improves the fitness function to ensure the safety of the path. Third,the principle of motion optimization is used to update points on the path to reduce the blindness of position updates. Simulation results show that the path obtained by the IBQL algorithm is excellent and using the improved beetle antennae search algorithm to optimize the path has little impact on the overall real-time performance. The optimized path has advantages such as fewer turning points and good smoothness.更多还原