Reproduce “CoLA: Multi-VLA Coordination via Learned Adapters” 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
Ahaskar Kashyap, Jiawen Zong, Boyu Ma, Chuhao Zhou, Jingliang Li, Stefano V. Albrecht, Jianfei Yang
CoLA enables robot arms to communicate without retraining pretrained vision-language-action backbones. It freezes each backbone, learns compact messages and new action heads from scripted demonstrations, and lets arms exchange messages directly. In one 150-episode simulated occluded tray task, success was 76.0% with messages and 1.3% with messages zeroed; this comparison also reflects handover failure.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.