Reproduce “FT-WBC: Learning Fault-Tolerant Whole-Body Control for Legged Loco-Manipulation” 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
Yudong Zhong, Pengfei Mai, Sikai Guo, Jiahang Cao, Zhihai Bi, Qiuyue Liu, Ziyan Feng, Jinni Zhou and 1 more
A framework called FT-WBC enables legged manipulators to perform loco-manipulation tasks robustly despite actuator failures by integrating a fault estimator and a posture adaptation module. This approach improves robot survival rates and maintains arm workspace under various failure conditions, demonstrating zero-shot sim-to-real transfer to a physical robot.
Key insights
- Explicitly detecting and localizing actuator failures at the joint level is crucial for low-level leg policies to synthesize effective compensatory gaits and maintain whole-body stability.
- High-level posture adaptation is necessary to arbitrate manipulation intentions with real-time fault conditions, preventing unstable base poses and effectively balancing stability with arm workspace preservation.
- A 'survival-first' design philosophy, where conservative posture adjustments prioritize preventing catastrophic falls over aggressive task completion under extreme faults, enhances safety and reliability.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.