Event
RISE: Guiding Humanoid Fall Recovery with Task Objectives at Inference Time
1 reports · 1 sources
Osher Azulay, Cheng-Lin Hsieh, and Stella X. Yu propose the RISE method, which enables a perceptive humanoid robot to use task objectives at inference time to guide the manner in which it stands up from a fall, without retraining. The method first learns a recovery vocabulary that associates intended actions with motion through sensory feedback; a visual diffusion planner then learns from executions of the vocabulary, predicting motion intent as future body states, and task costs guide the predictions toward a desired orientation, rising speed, or path, thereby generating motion latent variables. Simulations demonstrate diverse recoveries, terrain adaptation, and test-time guidance, and real-robot demonstrations were conducted on a Unitree G1.
Report timeline
- RISE: Test-Time Motion Steering for Perceptive Humanoid Fall Recovery
RISE is a perceptive humanoid fall-recovery method that uses task objectives at inference time to steer how the robot stands up. It learns a recovery vocabulary that links intended movements to motor actions through sensory feedback; a visual diffusion planner learns from executions of this vocabulary to predict motion intent as future body states, and task costs steer that prediction toward a desired heading, rise speed, or path, after which motion latents are generated accordingly. Simulations demonstrate diverse recoveries, terrain adaptation, and test-time steering without retraining, with a real-robot demo on the Unitree G1.
arXiv cs.RO ·