PTDL:Multi-Terrain Fall Recovery via Phase-Terrain Decoupled Learning

2026-06-07

AI Summary

This work addresses the challenge of enabling humanoid robots to seamlessly transition between fall recovery and walking across diverse unstructured terrains using only proprioceptive sensing. To this end, the authors propose a phase–terrain decoupled learning framework that trains a single proprioceptive policy by disentangling supervision signals for recovery versus walking phases and across different terrains during training. The approach incorporates a dual-motion prior discriminator gated by projected gravity, a ground-to-walking transition mechanism, and terrain-hierarchical recovery shaping, allowing implicit selection of appropriate recovery strategies at deployment without explicit terrain labels. Experiments on the Unitree G1 robot demonstrate stable recovery on flat ground, gravel, and 20° slopes, along with terrain-adaptive rising behaviors and seamless transitions from recovery to locomotion.

Abstract

Humanoid robots can fall on slopes, gravel, and uneven ground in unstructured environments. We target integrated fall recovery and locomotion: rebuilding balance from a fallen state using proprioception alone and resuming velocity-commanded walking at the fall site. Prior methods often stop at quasi-static rise, neglect the post-fall ground-contact phase, or, when trained on mixed terrains without separating recovery and locomotion phases or per-surface constraints, collapse to a single compromise get-up across surfaces. We propose Phase--Terrain Decoupled Learning (PTDL), which decouples training supervision along phase and terrain axes while deploying one proprioceptive policy. On the phase axis, projected-gravity-gated dual motion-prior discriminators and a probe-to-walk transition link post-fall recovery to commanded walking. On the terrain axis, terrain-stratified recovery shaping assigns surface-specific training supervision on flat ground, gravel, and slopes; terrain labels are training-only and withheld from policy observations, enabling implicit post-fall strategy selection at deployment. We validate PTDL on a 29-DoF Unitree G1 across flat ground, gravel, and slopes up to 20 degrees in simulation and on hardware, achieving stable cross-terrain recovery, smooth recovery-to-locomotion transitions, and differentiated post-fall rise behaviors under one deployed policy.

Problem

• fall recovery
• humanoid robots
• multi-terrain
• proprioception
• locomotion

Innovation

• Phase-Terrain Decoupled Learning
• fall recovery
• proprioceptive policy
• terrain-stratified training
• humanoid locomotion