Learning to Hop for a Single-Legged Robot with Parallel Mechanism

2025/01/21 by Zhang, Hongbo, Chu, Xiangyu, Chen, Yanlin +4 · 1 citation
#FOS: Computer and information sciences #Robotics (cs.RO)

paper · doi:10.48550/arxiv.2501.11945

Abstract

This work presents the application of reinforcement learning to improve the performance of a highly dynamic hopping system with a parallel mechanism. Unlike serial mechanisms, parallel mechanisms can not be accurately simulated due to the complexity of their kinematic constraints and closed-loop structures. Besides, learning to hop suffers from prolonged aerial phase and the sparse nature of the rewards. To address them, we propose a learning framework to encode long-history feedback to account for the under-actuation brought by the prolonged aerial phase. In the proposed framework, we also introduce a simplified serial configuration for the parallel design to avoid directly simulating parallel structure during the training. A torque-level conversion is designed to deal with the parallel-serial conversion to handle the sim-to-real issue. Simulation and hardware experiments have been conducted to validate this framework.

Cited by

• Bridging the Sim-to-Real Gap in Parallel-Link Leg Mechanisms via Simulator-Side Dynamics Normalization

Related

• Selecting Mechanical Parameters of a Monopode Jumping System with Reinforcement Learning
• Towards General and Autonomous Learning of Core Skills: A Case Study in Locomotion
• Robust and Versatile Bipedal Jumping Control through Reinforcement Learning
• An Open-Loop Baseline for Reinforcement Learning Locomotion Tasks
• Legged Robots that Keep on Learning: Fine-Tuning Locomotion Policies in the Real World