Active Learning of Discrete-Time Dynamics for Uncertainty-Aware Model Predictive Control

TL;DR

This paper introduces an active, self-supervised learning method for nonlinear robotic system dynamics that combines offline and online learning, enabling real-time, uncertainty-aware model predictive control adaptable to changing environments.

Contribution

It presents a novel adaptive learning framework that integrates offline and online data to improve model accuracy and control performance in real-time robotic applications.

Findings

High sample efficiency and real-time adaptation demonstrated on quadrotor experiments.

Outperforms classical and adaptive control baselines in diverse flight conditions.

Shows resilience and generalization to unseen environments.

Abstract

Model-based control requires an accurate model of the system dynamics for precisely and safely controlling the robot in complex and dynamic environments. Moreover, in the presence of variations in the operating conditions, the model should be continuously refined to compensate for dynamics changes. In this paper, we present a self-supervised learning approach that actively models the dynamics of nonlinear robotic systems. We combine offline learning from past experience and online learning from current robot interaction with the unknown environment. These two ingredients enable a highly sample-efficient and adaptive learning process, capable of accurately inferring model dynamics in real-time even in operating regimes that greatly differ from the training distribution. Moreover, we design an uncertainty-aware model predictive controller that is heuristically conditioned to the aleatoric (data) uncertainty of the learned dynamics. This controller actively chooses the optimal control actions that (i) optimize the control performance and (ii) improve the efficiency of online learning sample collection. We demonstrate the effectiveness of our method through a series of challenging real-world experiments using a quadrotor system. Our approach showcases high resilience and generalization capabilities by consistently adapting to unseen flight conditions, while it significantly outperforms classical and adaptive control baselines.