A single blind proprioceptive RL policy, trained with randomized terrain compliance and unevenness, lets the HRP-5P humanoid walk on real compliant and uneven terrain; a clock-modulation extension improves simulated robustness on uneven ground.
Terrain classification and locomotion parameters adaptation for humanoid robots using force/torque sensing,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Robust Humanoid Walking on Compliant and Uneven Terrain with Deep Reinforcement Learning
A single blind proprioceptive RL policy, trained with randomized terrain compliance and unevenness, lets the HRP-5P humanoid walk on real compliant and uneven terrain; a clock-modulation extension improves simulated robustness on uneven ground.