A world-action model conditioned on optical-flow history and kinematic descriptors improves dynamic manipulation success by 15.5 points on DOMINO and 22.9 points in real-world tests.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
DynamicWAM: Dual-Path Motion Conditioning for World-Action Models in Dynamic Manipulation
A world-action model conditioned on optical-flow history and kinematic descriptors improves dynamic manipulation success by 15.5 points on DOMINO and 22.9 points in real-world tests.