Automatic Differentiation and Continuous Sensitivity Analysis of Rigid Body Dynamics
This work addresses the challenge of physical understanding in robotics for improved model-based control, though it appears incremental as it builds on existing differentiable simulation techniques.
The authors tackled the problem of learning dynamics models for robots by introducing a differentiable physics simulator for rigid body dynamics, which enabled parameter estimation and a closed-loop model-predictive control algorithm that achieved cost-minimizing performance.
A key ingredient to achieving intelligent behavior is physical understanding that equips robots with the ability to reason about the effects of their actions in a dynamic environment. Several methods have been proposed to learn dynamics models from data that inform model-based control algorithms. While such learning-based approaches can model locally observed behaviors, they fail to generalize to more complex dynamics and under long time horizons. In this work, we introduce a differentiable physics simulator for rigid body dynamics. Leveraging various techniques for differential equation integration and gradient calculation, we compare different methods for parameter estimation that allow us to infer the simulation parameters that are relevant to estimation and control of physical systems. In the context of trajectory optimization, we introduce a closed-loop model-predictive control algorithm that infers the simulation parameters through experience while achieving cost-minimizing performance.