Single Model Controls Multiple Robots: An Analysis of the CHORUS VLA

DJiafei · x · 2026-07-29

Traditional multi-robot collaboration typically relies on complex centralized control, which scales poorly with team size. To address this, researchers introduced CHORUS.

By training a single Vision-Language-Action (VLA) model, CHORUS can control multiple diverse robots using only local observations and identifying information. This allows robots to collaborate on complex manipulation tasks such as using a tape measure, handing over a book, or lifting a laundry basket.

Related event: CHORUS: Decentralized Multi-Robot Collaboration via Single VLA Model(6 posts)→

Original post →

More from Embodied

Embodied channel →