机器人2025，Vol.47Issue(5)：625-635,11.DOI:10.13973/j.cnki.robot.240239
基于预训练—微调框架的四足机器人结构—控制协同设计
Structure-control Co-design of Quadruped Robots Based on Pre-training-Fine-tuning Framework
摘要
Abstract
In nature,animals with exceptional locomotion abilities,such as cougars,often possess asymmetric fore and hind legs,with their powerful hind legs acting as reservoirs of energy for leaps.Inspired by this biological concept,a co-optimization method for mechanical structures and control strategies is proposed,focusing on optimizing the leg length of the robot to enhance its overall motion performance.Firstly,a novel pretraining-finetuning framework is introduced,which not only guarantees optimal control strategies for each mechanical candidate but also improves the training efficiency of the algorithm.Additionally,spatial domain randomization is integrated with discount regularization,which remarkably improves the generalization ability of the pretraining network.Experimental results indicate that the proposed pretraining-finetuning framework significantly enhances the overall co-design performance with less time consumption.Moreover,the co-design strategy substantially exceeds the conventional method of independently optimizing control strategies,providing an innovative approach to enhancing the extreme parkour capabilities of quadruped robots.关键词
四足机器人/结构—控制协同设计/深度强化学习Key words
quadrupedal robot/structure-control co-design/deep reinforcement learning引用本文复制引用
陈词,余纪宇,李超,陆豪健,高洪波,熊蓉,王越..基于预训练—微调框架的四足机器人结构—控制协同设计[J].机器人,2025,47(5):625-635,11.基金项目
国家自然科学基金(62373322,62173293,62303407,U2013601). （62373322,62173293,62303407,U2013601）
