Reproduce “Modeling and Understanding Neural Weights of Humanoid Policies with Diffusion” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Huixuan Cheng, Bike Zhang, Shankar Sastry, Koushil Sreenath
The authors generate humanoid controllers by compressing policy weights into a latent space, learning their distribution with diffusion, and decoding samples into runnable networks. Separate unconditional models produce walking, dancing, and spin-kick policies for a Unitree G1, deployed without fine-tuning. For spin kick, diffusion outperforms matched-distance random perturbations, but evidence is limited to three motions on one robot.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.