Sample Efficient Learning of Body-Environment Interaction of an Under-Actuated System

TL;DR

This paper compares different methods for learning the motility map in under-actuated robotic systems interacting with high-friction environments, highlighting trade-offs between simplicity and complexity based on data availability.

Contribution

It introduces a systematic comparison of modeling approaches for learning the motility map from motion data of a physical robot with complex interactions.

Findings

Simpler methods perform better with small datasets.

More sophisticated methods excel with larger datasets.

Trade-off identified between model complexity and training data size.

Abstract

Geometric mechanics provides valuable insights into how biological and robotic systems use changes in shape to move by mechanically interacting with their environment. In high-friction environments it provides that the entire interaction is captured by the ``motility map''. Here we compare methods for learning the motility map from motion tracking data of a physical robot created specifically to test these methods by having under-actuated degrees of freedom and a hard to model interaction with its substrate. We compared four modeling approaches in terms of their ability to predict body velocity from shape change within the same gait, across gaits, and across speeds. Our results show a trade-off between simpler methods which are superior on small training datasets, and more sophisticated methods, which are superior when more training data is available.