Modeling and Understanding Neural Weights of Humanoid Policies with Diffusion
arXiv cs.AI · 15 Oct 2026, 03:59 AEST
Recent advances in robot learning have driven rapid progress in humanoid control, enabling increasingly agile behaviors in the real world. This work investigates whether high-performing humanoid control networks can be generatively modeled directly through their weights.
Key facts
- Humanoid control networks are typically obtained through conventional training procedures, while comparatively little attention has been given to directly modeling the distribution of their learned weights.
- We adopt a latent diffusion framework that couples a linear autoencoder with a Diffusion Transformer.
- Trained on networks collected from reinforcement learning for the Unitree G1 robot, our approach generates deployable networks that perform various motions in the real world, including walking, dancing, and spin kick.
- We identify a structural principle in the first-layer weights that consistently reflects the robot's kinematic prior.
Read aloud
Recent advances in robot learning have driven rapid progress in humanoid control, enabling increasingly agile behaviors in the real world. This work investigates whether high-performing humanoid control networks can be generatively modeled directly through their weights. Humanoid control networks are typically obtained through conventional training procedures, while comparatively little attention has been given to directly modeling the distribution of their learned weights. We adopt a latent diffusion framework that couples a linear autoencoder with a Diffusion Transformer. Trained on networks collected from reinforcement learning for the Unitree G1 robot, our approach generates deployable networks that perform various motions in the real world, including walking, dancing, and spin kick. We identify a structural principle in the first-layer weights that consistently reflects the robot's kinematic prior.
Sources in this cluster
- Modeling and Understanding Neural Weights of Humanoid Policies with Diffusion
arXiv cs.AI · 15 Oct 2026, 03:59 AEST