SuperMimic: Humanoid Control with One Differentiable Model

The paper presents SuperMimic, a method that uses a single differentiable world model to amortize the common articulated-body and contact dynamics in humanoid motion tracking, significantly reducing computational requirements by over 300-fold. This method surpasses a 3 orders of magnitude baseline in both accuracy and compute efficiency when optimizing complex sequences of motion. SuperMimic demonstrates the practical applicability of this approach in a variety of challenging human motions (e.g., flips, rolls, and jumps) involving extensive use of contact with the environment. The model achieves zero-errors for these complex maneuvers which represent different contact-rich scenarios proving to be more cost-effective and robust in leveraging a parallelizable differentiable simulated model for real controllers. Transferring training to a real-world humanoid robot showcases the sim-to-real efficacy of the approach Problem setting and central claim Humanoid whole-body motion tracking is commonly formulated as physics-based imitation: a policy receives a reference motion and produces joint-level actions that reproduce it under contact-rich rigid-body dynamics. The principal scaling challenge is not the existence of motion data, but the computational cost of optimizing one controller over a large and heterogeneous motion library. Existing systems either train motion-specific experts and distill them into a generalist policy or directly optimize a generalist policy with model-free RL. In both cases, simulator rollouts are repeatedly used to estimate action consequences, despite the fact that all motions are executed by the same humanoid and therefore share the same underlying dynamics. “SuperMimic: Scaling Humanoid Whole-Body Control with One Differentiable World Model” (2610.28480) addresses this bottleneck with a learned differentiable world model trained from simulator transitions. The central proposal is to amortize the common articulated-body and contact dynamics across the entire motion library, then optimize the tracking policy by backpropagation through short imagined rollouts. The authors report that this approach reaches or exceeds the tracking performance of the cluster-scale SONIC baseline with one GPU, while reducing the reported compute requirement by more than three orders of magnitude. The paper’s strongest claim is explicitly computational: a single learned world model removes compute as the dominant bottleneck for scaling humanoid whole-body …