ForPC force-aware whole-body control enables humanoid cart pushing

ZhenYu Zou and other authors propose ForPC, a force-aware whole-body control framework that regulates a humanoid robot's cart-pushing speed via two-stage reinforcement learning. The method infers force latent variables and state latent variables from proprioceptive observations, requiring no external force sensors, and then a residual policy coordinates arm, waist and leg movements. In simulation, cart-pushing speed RMSE is reduced by up to 68.2% compared with a locomotion-control-only baseline, and lateral error on multi-curve routes is reduced by up to 74.2%. The authors state that the policy was trained only in simulation and was transferred to a physical Unitree G1 without real-world training or fine-tuning, and without load-specific adjustment, pushing carts loaded with up to 30 kg at 0.5 m/s.

AI 根据报道生成 · 2 days ago更新