Reproduce “WT-UMI: Tactile-based Whole-Body Manipulation via Force-Supervised Contact-Aware Planning” 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
Jaehwi Jang, Zhaoyuan Gu, Alfred Cueva, Zimeng Chai, Junjie Sheng, Thong Nguyen, Himank Galundia, Yifan Wu and 10 more
A novel system enables humanoid robots to execute complex, contact-rich whole-body manipulation by integrating a wearable tactile interface (WT-UMI) with a learning framework that explicitly plans and regulates distributed contact forces. This approach achieves higher success rates and reduced contact-position errors for tasks involving deformable and bulky objects.
Key insights
- Human demonstrations, when combined with a data-efficient target-pose correction mechanism, provide high-quality, smoother force supervision for robot learning.
- Explicitly planning and regulating contact force trajectories, rather than relying on implicit contact handling, leads to more stable and robust whole-body manipulation.
- A shared wearable tactile interface across human demonstrators and robots effectively reduces the domain gap, improving the transferability of learned skills.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.