A Stability Training Method of Legged Robots Based on Training Platforms and Reinforcement Learning with Its Simulation and Experiment
Open Access
- 31 August 2022
- journal article
- research article
- Published by MDPI AG in Micromachines
- Vol. 13 (9) , 1436
Abstract
This paper continues the proposed idea of stability training for legged robots with any number of legs and any size on a motion platform and introduces the concept of a learning-based controller, the global self-stabilizer, to obtain a self-stabilization capability in robots. The overall structure of the global self-stabilizer is divided into three modules: action selection, adjustment calculation and joint motion mapping, with corresponding learning algorithms proposed for each module. Taking the human-sized biped robot, GoRoBoT-II, as an example, simulations and experiments in three kinds of motions were performed to validate the feasibility of the proposed idea. A well-designed training platform was used to perform composite random amplitude-limited disturbances, such as the sagittal and lateral tilt perturbations (±25°) and impact perturbations (0.47 times the robot gravity). The results show that the proposed global self-stabilizer converges after training and can dynamically combine actions according to the system state. Compared with the controllers used to generate the training data, the trained global self-stabilizer increases the success rate of stability verification simulations and experiments by more than 20% and 15%, respectively.Keywords
Funding Information
- National Key Research and Development Program of China (2018YFB1304502, 61936004)
- Major program of National natural science foundation of China (2018YFB1304502, 61936004)
This publication has 23 references indexed in Scilit:
- Dynamic Balance Optimization and Control of Quadruped Robot Systems With Flexible JointsIEEE Transactions on Systems, Man, and Cybernetics: Systems, 2015
- Humanoid Robot Balance Control Using the Spherical Inverted Pendulum ModeFrontiers in Robotics and AI, 2015
- The motirod: a novel physical skill task that enhances motivation to learn and thereby increases neurogenesis especially in the female hippocampusBrain Research, 2015
- Short-term motor learning of dynamic balance control in children with probable Developmental Coordination DisorderResearch in Developmental Disabilities, 2015
- Influence of Posturographic Platform Biofeedback Training on the Dynamic Balance of Adult Stroke PatientsJournal of Stroke and Cerebrovascular Diseases, 2014
- Robust feedback control of ZMP-based gait for the humanoid robot NaoThe International Journal of Robotics Research, 2013
- Real time implementation of CTRNN and BPTT algorithm to learn on-line biped robot balance: Experiments on the standing postureControl Engineering Practice, 2011
- Experimental Study of Humanoid Robot HRP-1SThe International Journal of Robotics Research, 2004
- Dynamic balance of a biped robot using fuzzy reinforcement learning agentsFuzzy Sets and Systems, 2003
- An Approach to Biped Control SynthesisRobotica, 1989