Developing robust locomotion controllers for bipedal robots with closed kinematic chains presents unique challenges, particularly since most reinforcement learning (RL) approaches simplify these parallel mechanisms into serial models during training. We demonstrate that this simplification significantly impairs sim-to-real transfer by failing to capture essential aspects such as joint coupling, friction dynamics, and motor-space control characteristics. In this work, we present an RL framework that explicitly incorporates closed-chain dynamics and validate it on our custom-built robot TopA. Our approach enhances policy robustness through symmetry-aware loss functions, adversarial training, and targeted network regularization. Experimental results demonstrate that our integrated approach achieves stable locomotion across diverse terrains, significantly outperforming methods based on simplified kinematic models.
@article{arxiv.2507.10164,
title = {Robust RL Control for Bipedal Locomotion with Closed Kinematic Chains},
author = {Egor Maslennikov and Eduard Zaliaev and Nikita Dudorov and Oleg Shamanin and Karanov Dmitry and Gleb Afanasev and Alexey Burkov and Egor Lygin and Simeon Nedelchev and Evgeny Ponomarev},
journal= {arXiv preprint arXiv:2507.10164},
year = {2025}
}