Reproduce “ELAN4D: Embodiment-Centric 4D Supervision for Vision-Language-Action Models via Plug-and-Play Adaptation” 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

Zeyuan He, Bowen Yang, Zhirui Fang, Keru Zhou, Lei Jiang, Jingjing Qian, Fan Mo, Junchi Yan and 4 more

ELAN4D trains a vision-language-action robot policy to predict future 3D motion of robot-body points, using forward kinematics for supervision rather than external scene tracking. Its track decoder and point inputs are removed at inference, while its control branch remains. On AgileX Piper, spatial-task success for the π0.5-based policy rose from 15% to 65%, a result limited to tested settings.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.