CHORUS: Single VLA Controls Multiple Robots for Decentralized Collaboration

DJiafei · x · 2026-08-05

CHORUS introduces a decentralized approach to multi-robot coordination. Moving away from complicated centralized control that struggles to scale, the research trains a single Vision-Language-Action (VLA) model to control multiple diverse robots.

Relying only on local observations and identifying information, the robots can collaborate on complex manipulation tasks like using a tape measure, handing over a book, or lifting a laundry basket.

Original post →

More from Embodied

Embodied channel →