Reproduce “RefCloth: Transferable Adversarial Tablecloths for Vision-Language-Action Models” 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
Shaozhen Liu, Zhiwei Li, Jiayang Sun, Shuangjun Du, Yi Zhong, Shenrui Ma, Yaozu Liu, Qi Li and 1 more
RefCloth tests if a patterned cloth can disrupt vision-language-action policies without changing objects or accessing camera streams. The authors optimize one texture on a policy and deploy it unchanged across tasks and target policies. In simulation, it reduces π₀.₅ success by a net 43.75 percentage points across four LIBERO suites versus a matched plain cloth; physical effectiveness remains untested.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.