ImitationFlow: Learning Deep Stable Stochastic Dynamic Systems by Normalizing Flows
This addresses the challenge of representing stable stochastic dynamics for robotics and control applications, though it appears incremental as an extension of existing Normalizing Flows.
The paper tackles the problem of learning complex globally stable stochastic nonlinear dynamics from demonstrated trajectories by introducing ImitationFlow, which extends Normalizing Flows to learn stable Stochastic Differential Equations. The result is a model that outperforms previous algorithms in representation accuracy, eliminates the Gaussian assumption on demonstrations, and is validated on standard datasets and a real robot experiment.
We introduce ImitationFlow, a novel Deep generative model that allows learning complex globally stable, stochastic, nonlinear dynamics. Our approach extends the Normalizing Flows framework to learn stable Stochastic Differential Equations. We prove the Lyapunov stability for a class of Stochastic Differential Equations and we propose a learning algorithm to learn them from a set of demonstrated trajectories. Our model extends the set of stable dynamical systems that can be represented by state-of-the-art approaches, eliminates the Gaussian assumption on the demonstrations, and outperforms the previous algorithms in terms of representation accuracy. We show the effectiveness of our method with both standard datasets and a real robot experiment.