CoLA: Multi-VLA Coordination via Learned Adapters
CoLA: Multi-VLA Coordination via Learned Adapters
This research proposes CoLA, a coordination framework for multiple VLA agents that maintains frozen VLA backbones and uses a lightweight message channel and diffusion action head for decentralized execution. The method achieves significant performance improvements in occluded tray handover tasks, with 76.0% success compared to 1.3% when messages are disabled. The study also highlights the importance of message channels in scenarios with private information.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
each agent's VLA backbone frozen and trains a lightweight message channel and a per-arm diffusion action head on top, with fully decentralised execution. Across three tasks that differ in how much each agent can observe, the channel matters only when information is private. On an occluded tray handover, where the box's colour marker is visible to one agent only, CoLA reaches 76.0% success versus 1.3% with messages ze
Open source S3
roed throughout training. On two-arm and three-arm handover tasks, removing messages has no measurable effect with a shared overhead camera (90.6% vs. 88.3% at two arms; 63.5% vs. 60.0% at three), but lowers success on the two-arm task's wrist-only variant (74.8% vs. 64.3%). This architecture generalises across different VLA backbones as well. With a frozen $\pi_{0.5}$ backbone on the same occluded tray task, CoLA ma
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org