The invention discloses a hexapod robot fault-tolerant gait planning method based on reinforcement learning. The hexapod robot fault-tolerant gait planning method comprises the following steps that a simulation model of a hexapod robot is built; establishing a CPG gait generator based on an improved Hopf oscillator; a simulation model, a strategy network and a CPG gait generator based on an improved Hopf oscillator are fused, a reinforcement learning motion control framework is constructed, and the reinforcement learning motion control framework is fused for the established hexapod robot simulation model; simulating that part of legs of the hexapod robot break down, freezing the broken-down legs of the hexapod robot in a simulation environment, and training the reinforcement learning motion control framework; the trained control network is integrated in a gait control framework of the hexapod robot and used for generating and verifying fault-tolerant gaits of the hexapod robot, if autonomous movement can be completed under the condition that the faulty legs are frozen, it is indicated that the strategy network is effective, and therefore the strategy network is extracted to be used for controlling real hexapod robot movement.
本发明公开了一种基于强化学习的六足机器人容错步态规划方法,包括以下步骤:搭建六足机器人的仿真模型;建立基于改进的Hopf振荡器的CPG步态生成器;融合仿真模型、策略网络与基于改进的Hopf振荡器的CPG步态生成器,构建强化学习运动控制框架,对于建立的六足机器人仿真模型,融合强化学习运动控制框架;模拟六足机器人部分腿出现故障,在仿真环境中冻结六足机器人出现故障的腿,对强化学习运动控制框架进行训练;将训练后的控制网络集成在六足机器人的步态控制框架中,用于生成六足机器人的容错步态并验证,如果能在冻结故障腿的情况下完成自主运动则表示策略网络有效,从而将策略网络提取出来用于控制真实的六足机器人运动。
Hexapod robot fault-tolerant gait planning method based on reinforcement learning
一种基于强化学习的六足机器人容错步态规划方法
2023-07-18
Patent
Electronic Resource
Chinese
Free fault-tolerant gait planning method and device for hexapod robot and storage medium
European Patent Office | 2020
|Double-gait hexapod robot and double-gait hexapod robot crawling method
European Patent Office | 2023
|Fault-tolerant motion planning method for hexapod robot with joint locking fault
European Patent Office | 2020
|