Reproduce “SAC-Loco: Safe and Adjustable Compliant Quadrupedal Locomotion” 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
Aoqian Zhang, Zixuan Zhuang, Chunzheng Wang, Shuzhi Sam Ge, Fan Shi, Cheng Xiang
Researchers at the National University of Singapore and Sun Yat-sen University introduced SAC-Loco, a safety-aware framework for quadrupedal robots that integrates adjustable disturbance compliance with robust failure prevention. The system demonstrated higher success rates under external forces up to 600N, maintaining stability during intentional dragging attempts that caused baselines to fail at 120.4N and 193.8N peak forces.
Key insights
- Integrating specialized policies for compliant locomotion and robust recovery, coordinated by a learned safety critic, allows for superior performance across a wide range of disturbance magnitudes compared to single-policy or fixed-compliance approaches.
- The teacher-student distillation framework effectively transfers complex behaviors learned with privileged information in simulation to policies deployable on physical hardware using only proprioceptive sensors, eliminating the need for explicit force sensing.
- Leveraging Corrected Capture Point (CCP) dynamics in the safe policy significantly enhances a robot's ability to actively restore balance and prevent failure under large, destabilizing forces.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.