摘要
为解决稀疏立足地形中单腿失效六足机器人无法继续稳定行走的问题,提出一种失稳躲避蒙特卡洛树搜索算法(IAMCTS)用于容错步态规划。将容错步态规划问题转化为马尔科夫序列优化过程;根据稳定性判据制定了失稳状态躲避策略,并将其引入蒙特卡洛树搜索算法模拟阶段动作选择策略中,避免选择失稳支撑动作以生成容错步态序列。仿真系统环境中进行模拟对比分析表明,所提规划算法具有一定可行性和优势性;在IAMCTS规划的自由容错步态模式下,单腿失效六足机器人在稀疏立足地形中的平均前进距离增长了88.45 cm,平均前进速度提高了18.91%。
To solve the problem that single-leg failed hexapod robot cannot continue to walk steadily in sparse terrain,a instability avoid Monte Carlo tree search algorithm(IAMCTS)is proposed for fault-tolerant gait planning.The faulttolerant gait planning problem is transformed into a Markov sequence optimization process.According to the stability criteria,the unstable state avoidance strategy is formulated,and it is introduced into the action selection strategy of the Monte Carlo tree search algorithm during simulation phase to generate a fault-tolerant gait sequence by avoiding the unstable support state.The comparative analysis in the simulation system environment shows that the proposed planning method has certain feasibility and advantages.Under the free fault-tolerant gait mode planned by IAMCTS,the average forward distance increased by 88.45 cm and the average forward speed increased by 18.91%of the single-leg failed hexapod robot in sparse terrain.
作者
刘恒佳
胡立坤
徐大也
王小勇
Liu Hengjia;Hu Likun;Xu Daye;Wang Xiaoyong(School of Electric Engineering,Guangxi University,Nanning 530004,China)
出处
《国外电子测量技术》
北大核心
2023年第8期154-161,共8页
Foreign Electronic Measurement Technology
基金
广西科技计划项目(桂科AB21220039)资助。
关键词
六足机器人
容错步态
强化学习
蒙特卡洛树搜索
hexapod robot
fault-tolerant gait
reinforcement learning
Monte Carlo tree search
