Tracking large human-motion datasets has become the dominant route to general, expressive humanoid whole-body control, but scaling across many motions is expensive. Current pipelines use model-free reinforcement learning, either merging per-motion specialists through lossy distillation, or brute-forcing a single policy with cluster-scale compute. SuperMimic amortizes learning across all motions with one differentiable world model of their shared dynamics, fit from logged transitions of a standard rigid-body simulator. The world model and the tracking policy train together in one loop, and backpropagation-through-time over short imagined rollouts gives the policy analytic gradients. Fast imagined rollouts and analytic gradients make policy learning efficient. In 3 hours on a single GPU, SuperMimic surpasses the return of a state-of-the-art motion tracker baseline trained for 7 days on 128 GPUs. It reduces GPU-hours by more than $1000\times$ and fails on only 25 of 129,514 benchmark motions, versus 179 failures for the baseline. After 9.8 hours on two GPUs, SuperMimic achieves zero failures. From 1 to 264 hours of motion, the world model's prediction error stays stable, and GPU-hours to a target return grow slowly. Our policy transfers to a physical Unitree G1 across diverse and acrobatic motions. SuperMimic removes compute as the bottleneck to scaling humanoid whole-body control, making it practical to measure motion-tracking scaling laws.
Learning Differentiable Reachability Maps for Optimization-based Humanoid Motion Generation
CNRS-AIST JRL (Joint Robotics Laboratory)(法国国家科学研究中心-日本产业技术综合研究所联合机器人实验室);IRL(智能机器人实验室);National Institute of Advanced Industrial Science and Technology (AIST)(国家先进工业科学与技术研究所)
HumanoidNDM: Closing the Training-Deployment Gap of Motion Tracking via Humanoid Neural Dynamics Model
Peking University(北京大学);Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所);Beijing Academy of Artificial Intelligence(北京智源人工智能研究院);Tsinghua University(清华大学)
OneSteer: Unlocking Steerable General Humanoid Motion Tracking via One-Step Diffusion Whole-Body Control
Southeast University(东南大学);Southern University of Science and Technology(南方科技大学);The Hong Kong Polytechnic University(香港理工大学);Shenzhen University of Advanced Technology(深圳先进技术大学)
SemPhys: Semantic-Physical Co-Learning for Humanoid Motion Generation via Differentiable World Models
Zhejiang University(浙江大学);Westlake University(西湖大学);Shanghai Jiao Tong University(上海交通大学);Shanghai Innovation Institute(上海创新研究院);Beijing Institute of Technology(北京理工大学);Huazhong University of Science and Technology(华中科技大学);Northwest Polytechnical University(西北工业大学);Xmov(星动科技);Fudan University(复旦大学)
Differentiable Information Enhanced Model-Based Reinforcement Learning
Institute for Artificial Intelligence, Peking University(北京大学人工智能研究院);State Key Laboratory of General Artificial Intelligence, Peking University(北京大学通用人工智能国家重点实验室);State Key Laboratory of General Artificial Intelligence, BIGAI(北京通用人工智能研究院通用人工智能国家重点实验室);Institute of automation, Chinese Academy of Sciences(中国科学院自动化研究所)