Reproduce “AffectiveVLA: Safe and Intent-Aware VLA Control for Human-Humanoid Social Interaction” 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
Nikita Kuzmin, Georgii Demianchuk, Mariya Lezina, Yara Mahmoud, Iana Zhura, Miguel Altamirano Cabrera, Dzmitry Tsetserukou
AffectiveVLA adds a two-camera intent check before a vision-language-action policy, prompting it after handshake, fist-bump, or hug intent passes a confidence threshold. On a robot, adding the system to GR00T raised macro-F1 from 0.75 to 0.90, with fist-bump F1 rising from 0.50 to 0.78; the comparison does not isolate the threshold’s effect or establish safety.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.