Sim-to-Real of Soft Robots with Learned Residual Physics

Junpeng Gao*, Mike Yan Michelis*, Andrew Spielberg, Robert K. Katzschmann
ETH Zürich
IEEE Robotics and Automation Letters 2024
*Indicates Equal Contribution
Paper Code arXiv

We propose a new hybrid learning framework that combines analytical simulation with data-driven residual physics to model soft robots with large degrees of freedom. Our method bridges the sim-to-real gap and outperforms traditional system identification and pure data-driven methods.

Abstract

Accurately modeling soft robots in simulation is computationally expensive and commonly falls short of representing the real world. This well-known discrepancy, known as the sim-to-real gap, can have several causes, such as coarsely approximated geometry and material models, manufacturing defects, viscoelasticity and plasticity, and hysteresis effects. Residual physics networks learn from real-world data to augment a discrepant model and bring it closer to reality. Here, we present a residual physics method for modeling soft robots with large degrees of freedom. We train neural networks to learn a residual term -- the modeling error between simulated and physical systems. Concretely, the residual term is a force applied on the whole simulated mesh, while real position data is collected with only sparse motion markers. The physical prior of the analytical simulation provides a starting point for the residual network, and the combined model is more informed than if physics were learned tabula rasa. We demonstrate our method on 1) a silicone elastomeric beam and 2) a soft pneumatic arm with hard-to-model, anisotropic fiber reinforcements. Our method outperforms traditional system identification up to 60%. We show that residual physics need not be limited to low degrees of freedom but can effectively bridge the sim-to-real gap for high dimensional systems.

Overall pipeline

In this paper, We provide a full pipeline (as shown below) from spatially sparse data to dense residual force estimation and apply our approach to both software and hardware experiments. Our approach is easy to apply in practice and can improve soft robotics engineering workflows through more reliable modeling.

overview pipeline
Overview of the residual physics pipeline for high dimensional systems, demonstrated with a soft robotic arm. The learned residual force compensates for state-to-state prediction errors, such that sparse motion markers in simulation match those in reality.

BibTeX

@article{gao2024sim,
        title={Sim-to-Real of Soft Robots with Learned Residual Physics},
        author={Gao, Junpeng and Michelis, Mike Yan and Spielberg, Andrew and Katzschmann, Robert K},
        journal={arXiv preprint arXiv:2402.01086},
        year={2024}
      }