Structured learning of rigid-body dynamics: A survey and unified view from a robotics perspective

TL;DR

This paper surveys methods that integrate analytical rigid-body mechanics with data-driven models to improve physical accuracy and data efficiency in modeling mechanical systems, providing a unified framework and discussing key techniques.

Contribution

It offers a comprehensive survey and unified view of combining data-driven regression models with analytical rigid-body mechanics, highlighting structural knowledge incorporation.

Findings

Unified framework for data-driven and analytical models

Analysis of latent functions and operators in rigid-body mechanics

Discussion of structured model design techniques like automatic differentiation

Abstract

Accurate models of mechanical system dynamics are often critical for model-based control and reinforcement learning. Fully data-driven dynamics models promise to ease the process of modeling and analysis, but require considerable amounts of data for training and often do not generalize well to unseen parts of the state space. Combining data-driven modelling with prior analytical knowledge is an attractive alternative as the inclusion of structural knowledge into a regression model improves the model's data efficiency and physical integrity. In this article, we survey supervised regression models that combine rigid-body mechanics with data-driven modelling techniques. We analyze the different latent functions (such as kinetic energy or dissipative forces) and operators (such as differential operators and projection matrices) underlying common descriptions of rigid-body mechanics. Based on this analysis, we provide a unified view on the combination of data-driven regression models, such as neural networks and Gaussian processes, with analytical model priors. Further, we review and discuss key techniques for designing structured models such as automatic differentiation.