Advancing Humanoid Locomotion: Mastering Challenging Terrains with Denoising World Model Learning
This addresses the challenge of enabling humanoid robots to navigate complex environments for applications in human-centric settings, representing a significant advance rather than an incremental improvement.
The paper tackled the problem of humanoid robot locomotion on challenging real-world terrains by introducing Denoising World Model Learning (DWL), an end-to-end reinforcement learning framework, resulting in the world's first humanoid robot mastering terrains like snow, stairs, and uneven ground with zero-shot sim-to-real transfer.
Humanoid robots, with their human-like skeletal structure, are especially suited for tasks in human-centric environments. However, this structure is accompanied by additional challenges in locomotion controller design, especially in complex real-world environments. As a result, existing humanoid robots are limited to relatively simple terrains, either with model-based control or model-free reinforcement learning. In this work, we introduce Denoising World Model Learning (DWL), an end-to-end reinforcement learning framework for humanoid locomotion control, which demonstrates the world's first humanoid robot to master real-world challenging terrains such as snowy and inclined land in the wild, up and down stairs, and extremely uneven terrains. All scenarios run the same learned neural network with zero-shot sim-to-real transfer, indicating the superior robustness and generalization capability of the proposed method.