Reproduce “DKWBC: Decoupling Motion Tracking and Balancing for Humanoid Whole-Body Control” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dingkun Wei, Zehong Shen, Huaijin Pi, Xueyang Wang, Xing Zhu, Hujun Bao, Yujun Shen, Xiaowei Zhou
DKWBC separates humanoid motion tracking from balance by computing nominal tracking torques with inverse dynamics and joint PD feedback, then learning residual torques for balanced execution. In a no-fine-tuning transfer test on BONES-SEED test-content under disturbance I, it reports 92.18% success versus 90.30% for BeyondMimic, with lower MPJPE and MPJVE. This result is limited to that test condition.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.