Qianzhong Chen*1, Junheng Li*2, Sheng Cheng3, Naira Hovakimyan3, and Quan Nguyen2
1 Stanford University, 2 University of Southern California, 3University of Illinois Urbana-Champaign
* Equal contributions
Bipedal locomotion control is essential for humanoid robots to navigate complex, human-centric environments. While optimization-based control designs are popular for integrating sophisticated models of humanoid robots, they often require labor-intensive manual tuning. In this work, we address the challenges of parameter selection in bipedal locomotion control using DiffTune, a model-based autotuning method that leverages differential programming for efficient parameter learning. A major difficulty lies in balancing model fidelity with differentiability. We address this difficulty using a low-fidelity model for differentiability, enhanced by a Ground Reaction Force-and-Moment Network (GRFM-Net) to capture discrepancies between MPC commands and actual control effects. We validate the parameters learned by DiffTune with GRFM-Net in hardware experiments, which demonstrates the parameters' optimality in a multi-objective setting compared with baseline parameters, reducing the total loss by up to 40.5% compared with the expert-tuned parameters. The results confirm the GRFM-Net's effectiveness in mitigating the sim-to-real gap, improving the transferability of simulation-learned parameters to real hardware.
Preprint: arXiv
We apply DiffTune to learn model-based locomotion MPC’s parameters for bipedal walking.
We propose GRFM-Net, a data-driven method to enhance the fidelity of a differentiable low-fidelity simulator, which reduces the sim-to-real gap when deploying the simulation-learned parameters to a real physical system.
Bibtex:
@misc{chen2024autotuningbipedallocomotionmpc,
title={Autotuning Bipedal Locomotion MPC with GRFM-Net for Efficient Sim-to-Real Transfer},
author={Qianzhong Chen and Junheng Li and Sheng Cheng and Naira Hovakimyan and Quan Nguyen},
year={2024},
eprint={2409.15710},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2409.15710},
}