Reproduce “Ask-to-Clarify: Resolving Instruction Ambiguity through Multi-turn Dialogue” 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
Xingyao Lin, Xinghao Zhu, Tianyi Lu, Guojin Zhong, Sicheng Xie, Hui Zhang, Xipeng Qiu, Zuxuan Wu and 1 more
Ask-to-Clarify connects dialogue with low-level visuomotor control, letting a robot resolve ambiguous instructions before acting. A vision-language planner handles clarification, while a separate diffusion-based executor generates motion through an adapter; the planner remains frozen during motor training. Across 11 tabletop tasks on an xArm 7, it achieves 82.7% Task Success Rate versus 78.2% for LLM-Policy, a descriptive result limited to this evaluation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.