Unlocks full-body musculoskeletal humanoid training by achieving order-of-magnitude speedups via massively parallel GPU simulation.
March 27, 2026
Original Paper
Towards Embodied AI with MuscleMimic: Unlocking full-body musculoskeletal motor learning at scale
arXiv · 2603.25544
The Takeaway
It democratizes biomechanically realistic motor learning, allowing a generalist policy to be trained on hundreds of human motions in days rather than months. This bridges the gap between simple robotic control and physiologically accurate human simulation.
From the abstract
Learning motor control for muscle-driven musculoskeletal models is hindered by the computational cost of biomechanically accurate simulation and the scarcity of validated, open full-body models. Here we present MuscleMimic, an open-source framework for scalable motion imitation learning with physiologically realistic, muscle-actuated humanoids. MuscleMimic provides two validated musculoskeletal embodiments - a fixed-root upper-body model (126 muscles) for bimanual manipulation and a full-body mo