The human foot serves as the critical interface between the body and environment during locomotion. We developed a novel contact-rich and deformable model of the human foot integrated within a complete musculoskeletal system MS-Human-700 that captures complex biomechanical interactions during walking, enabling stable and realistic locomotion control.
The foot-ground interface and skeleton-foot interface are strengthened by imposing elaborated constrains on the edges and vertices of the mesh. By emulating the compliant attachment in humans, we achieve a moderate balance between computational cost and simulation accuracy, incorporating 223 vertices and 426 triangular elements per foot.
We proposed a two-stage learning strategy to overcome the challenges posed by complex contact dynamics. In the first stage, we trained an initial policy on the rigid musculoskeletal model with simplified contact dynamics. And in the second stage, we transferred and fine-tuned this policy on the interface-enhanced model.
Benefit from the deformable foot, the interface-enhanced musculoskeletal model enables larger contact area and more balanced force distribution. This time-dependent distribution of the ground reaction force could improve stability in locomotion.
Human experiments demonstrate that the interface-enhanced musculoskeletal model can reproduce the human walking motion more closely, both in kinematics and force distribution.
This study addresses a long-standing limitation in musculoskeletal modeling by integrating a contact-rich, deformable
foot model into a conventional rigid musculoskeletal model and achieves:
(1) Physiologically accurate distribution and continuity of ground reaction forces,
(2) Significantly improved motion stability in velocity and acceleration variation,
(3) Biomechanically realistic walking patterns throughout the gait cycle that closely mirror human locomotion.
@inproceedings{gong2025contact,
title={Contact-Rich and Deformable Foot Modeling for Locomotion Control of the Human Musculoskeletal System},
author={Gong, Haixin and Zhang, Chen and Sui, Yanan},
booktitle={2025 IEEE-RAS 24th International Conference on Humanoid Robots (Humanoids)},
pages={301--308},
year={2025},
organization={IEEE}
}