Learning Force-Reactive Humanoid Loco-Manipulation
arXiv cs.AI · 27 Oct 2026, 03:59 AEST
A new framework called FLORE has been developed for learning force-reactive humanoid loco-manipulation. It successfully coordinates hand and body motion to control contact force in various tasks.
Key facts
- Many everyday tasks require controlling contact force as well as motion.
- A force-reactive diffusion policy takes vision, proprioception, and measured wrenches as input, then predicts actions and per-hand force commands.
- In simulation, the controller tracks per-hand force commands of varied magnitude and direction, with little loss in whole-body motion accuracy.
- Across three contact-rich tasks on the robot, the force-reactive policy succeeds in 80% of rollouts, compared with 26.7% for a position-only policy.
Read aloud
A new framework called FLORE has been developed for learning force-reactive humanoid loco-manipulation. It successfully coordinates hand and body motion to control contact force in various tasks. Many everyday tasks require controlling contact force as well as motion. A force-reactive diffusion policy takes vision, proprioception, and measured wrenches as input, then predicts actions and per-hand force commands. In simulation, the controller tracks per-hand force commands of varied magnitude and direction, with little loss in whole-body motion accuracy. Across three contact-rich tasks on the robot, the force-reactive policy succeeds in eighty percent of rollouts, compared with twenty six point seven percent for a position-only policy.
Sources in this cluster
- Learning Force-Reactive Humanoid Loco-Manipulation
arXiv cs.AI · 27 Oct 2026, 03:59 AEST