Wiggle and Go: robot learns rope dynamics from a single interaction at CoRL 2026

Arthur_K_J · x · 2026-09-20

A CoRL 2026 paper, Wiggle and Go, teaches robots to adapt to ambiguous dynamics the way humans learn: by interacting with the world. The robot wiggles a rope once, figures out how it behaves, then executes complex dynamic tasks using the estimated dynamics — with no real-world training data and no retries.

The authors present it as evidence that specialized world models still have a place, responding to Vincent Sitzmann's ECCV claim that only a video model could identify his cable's properties; they concede a VWM would be an exciting next test for Wiggle-based system identification.

Related event: Robot Learns Rope Dynamics with a Single Wiggle, Sparking Debate(3 posts)→

Original post →

More from Embodied

Embodied channel →