摘要
针对Q-Learning算法学习效率低、收敛速度慢且在动态障碍物的环境下路径规划效果不佳的问题,本文提出一种改进Q-Learning的移动机器人路径规划算法.针对该问题,算法根据概率的突变性引入探索因子来平衡探索和利用以加快学习效率;通过在更新函数中设计深度学习因子以保证算法探索概率;融合遗传算法,避免陷入局部路径最优同时按阶段探索最优迭代步长次数,以减少动态地图探索重复率;最后提取输出的最优路径关键节点采用贝塞尔曲线进行平滑处理,进一步保证路径平滑度和可行性.实验通过栅格法构建地图,对比实验结果表明,改进后的算法效率相较于传统算法在迭代次数和路径上均有较大优化,且能够较好的实现动态地图下的路径规划,进一步验证所提方法的有效性和实用性.
Aiming at the problems of low learning efficiency,slow convergence speed,and poor path planning effect in the environment of dynamic obstacles of the Q-Learning algorithm,an improved Q-Learning path planning algorithm for mobile robots is proposed.In an effort to deal with this problem,the algorithm introduces the exploration factor according to the mutation of probability to balance exploration and utilization to accelerate learning efficiency;and the exploration probability of the algorithm is guaranteed by designing deep learning factors in the update function;then incorporating the genetic algorithm to avoid getting into the local path optimum while exploring the optimal number of iteration steps by stage to reduce the repetition rate of dynamic map exploration;finally,the key node of the optimal path extracted by the output is smoothed by the Bessel curves curve to further ensure the smoothness and feasibility of the path.The experimental results show that the improved algorithm is more efficient than the traditional algorithm in terms of the number of iterations and paths,and it can achieve better path planning under dynamic maps,which further verifies the effectiveness and practicality of the proposed method.
作者
宋丽君
周紫瑜
李云龙
侯佳杰
何星
SONG Lijun;ZHOU Ziyu;LI Yunlong;HOU Jiajie;HE Xing(School of Information and Control Engineering,Xi′an University of Architecture and Technology,Xi′an 710055,China)
出处
《小型微型计算机系统》
CSCD
北大核心
2024年第4期823-829,共7页
Journal of Chinese Computer Systems
基金
中央军委装备发展部装备重大基础研究课题项目(2018921007)资助
国防科技重点实验室基金项目(2021-JCJQ-LB-070-08)资助.
