Thor: Towards Human-Level Whole-Body Reactions for Intense Contact-Rich Environments

2025-10-30

AI Summary

This study addresses the challenge of maintaining whole-body stability for humanoid robots under high-intensity, multi-contact disturbances. We propose a force-feedback-driven adaptive control framework. Methodologically, we (1) design a Force-Adaptive Trunk Tilt (FAT²) reward function that explicitly models the coupling between ground reaction forces and trunk orientation; and (2) develop a hierarchical reinforcement learning architecture with upper–lower body decoupling and coordinated waist parameter adaptation, unifying global perception and local response. The approach integrates force analysis, modular control, and end-to-end training, and is deployed on the Unitree G1 platform. Results demonstrate significant improvements: backward/forward pull resistance reaches 167.7 N / 145.5 N—approximately 70% higher than baseline methods. Furthermore, the robot successfully executes demanding anthropomorphic tasks—including pulling a cargo cabinet and opening a fire door with one hand—validating its whole-body dynamic adaptability and robustness under complex physical interactions.

Technology Category

• Intelligent Robots: Manipulation
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Humanoids hold great potential for service, industrial, and rescue applications, in which robots must sustain whole-body stability while performing intense, contact-rich interactions with the environment. However, enabling humanoids to generate human-like, adaptive responses under such conditions remains a major challenge. To address this, we propose Thor, a humanoid framework for human-level whole-body reactions in contact-rich environments. Based on the robot's force analysis, we design a force-adaptive torso-tilt (FAT2) reward function to encourage humanoids to exhibit human-like responses during force-interaction tasks. To mitigate the high-dimensional challenges of humanoid control, Thor introduces a reinforcement learning architecture that decouples the upper body, waist, and lower body. Each component shares global observations of the whole body and jointly updates its parameters. Finally, we deploy Thor on the Unitree G1, and it substantially outperforms baselines in force-interaction tasks. Specifically, the robot achieves a peak pulling force of 167.7 N (approximately 48% of the G1's body weight) when moving backward and 145.5 N when moving forward, representing improvements of 68.9% and 74.7%, respectively, compared with the best-performing baseline. Moreover, Thor is capable of pulling a loaded rack (130 N) and opening a fire door with one hand (60 N). These results highlight Thor's effectiveness in enhancing humanoid force-interaction capabilities.

Problem

• Enabling humanoids to generate human-like adaptive responses in contact-rich environments
• Addressing high-dimensional control challenges for humanoid whole-body stability
• Improving humanoid force-interaction capabilities during intense physical tasks

Innovation

• Force-adaptive torso-tilt reward function for human-like responses
• Reinforcement learning architecture decouples upper, waist, lower body
• Shared global observations and joint parameter updates across components