摘要
为提高未知环境下移动机器人自主导航性能,提出了一种未知环境下目标驱动的自主导航方法。不依赖实时更新的地图信息,移动机器人直接从激光雷达读数中构建候选目标点,并根据机器人自身位置和终点位置从候选目标点中选择最优点作为局部目标点。改进的基于深度强化学习的局部规划器从输入信息中直接输出动作信号,实现端到端的控制,使得移动机器人能够快速、安全地到达局部目标点,直至到达终点。实验表明,所提导航方法能够在未知复杂的环境下,可靠且高效地完成导航任务。与最近边界导航方法相比,平均路径长度减少了6.63%,平均运行时间缩短了19.11%,具有成功率高、路径短、速度快等优点。
In order to improve the autonomous navigation performance of mobile robots in unknown environment,a target-driven autonomous navigation method is proposed.Without relying on prior map information,the mobile robot directly constructs candidate target points from the LiDAR readings,and selects the best points from the candidate target points as the local target points according to the robot's own position and the goal position.The improved local planner based on deep reinforcement learning directly obtains action signals from input information to achieve end-to-end control,enabling the mobile robot to reach local target points quickly and safely until it reaches the goal.Experiments show that the proposed navigation method can accomplish navigation tasks reliably and efficiently under complex environments.Compared with the nearest frontier navigation method,the average path length is reduced by 6.63%,the average running time is shortened by 19.11%,and the method has the advantages of high success rate,short path and fast speed.
作者
徐建华
吴晓晖
张嘉轩
张钰荣
XU Jianhua;WU Xiaohui;ZHANG Jiaxuan;ZHANG Yurong(School of Automation,Beijing Institute of Technology,Beijing 100081,China)
出处
《中国惯性技术学报》
EI
CSCD
北大核心
2024年第3期250-257,共8页
Journal of Chinese Inertial Technology
基金
装备重大基础研究项目(5140502A03)。
关键词
未知环境
移动机器人
自主导航
深度强化学习
unknown environment
mobile robot
autonomous navigation
deep reinforcement learning
作者简介
徐建华(1975-),男,副研究员,博士,从事智能无人系统与智能避障研究。
