Whole-Body Model-Predictive Control of Legged Robots with MuJoCo

1Carnegie Mellon University, 2Google DeepMind
under review

Abstract

We demonstrate the surprising real-world effectiveness of a very simple approach to whole-body model-predictive control (MPC) of quadruped and humanoid robots: the iterative LQR (iLQR) algorithm with MuJoCo dynamics and finite-difference approximated derivatives. Building upon the previous success of model-based behavior synthesis and control of locomotion and manipulation tasks with MuJoCo in simulation, we show that these policies can easily generalize to the real world with few sim-to-real considerations. Our baseline method achieves real-time whole-body MPC on a variety of hardware experiments, including dynamic quadruped locomotion, quadruped walking on two legs, and full-sized humanoid bipedal locomotion. We hope this easy-to-reproduce hardware baseline lowers the barrier to entry for real-world whole-body MPC research and contributes to accelerating research velocity in the community.

Interactive GUI for real world legged robots.

quadruped walking on two legs

quadruped getting up on front legs

Go2 robot

full-sized humanoid robot trotting

BibTeX

@misc{zhang2025wholebodymodelpredictivecontrollegged,
        title={Whole-Body Model-Predictive Control of Legged Robots with MuJoCo}, 
        author={John Z. Zhang and Taylor A. Howell and Zeji Yi and Chaoyi Pan and Guanya Shi and Guannan Qu and Tom Erez and Yuval Tassa and Zachary Manchester},
        year={2025},
        eprint={2503.04613},
        archivePrefix={arXiv},
        primaryClass={cs.RO},
        url={https://arxiv.org/abs/2503.04613}, 
  }