Adaptive-Control-Oriented Meta-Learning for Nonlinear Systems

TL;DR

This paper introduces a control-oriented meta-learning approach that trains adaptive controllers in simulation to improve real-time nonlinear system control, demonstrated on a rotorcraft with wind disturbances.

Contribution

It proposes a novel meta-learning method focusing on control performance in closed-loop simulation, enhancing adaptive control for nonlinear systems with unknown dynamics.

Findings

Meta-learned controllers outperform regression-based methods in trajectory tracking.

The approach improves real-time adaptation in complex, dynamic environments.

Demonstrated effectiveness on a rotorcraft with wind disturbances.

Abstract

Real-time adaptation is imperative to the control of robots operating in complex, dynamic environments. Adaptive control laws can endow even nonlinear systems with good trajectory tracking performance, provided that any uncertain dynamics terms are linearly parameterizable with known nonlinear features. However, it is often difficult to specify such features a priori, such as for aerodynamic disturbances on rotorcraft or interaction forces between a manipulator arm and various objects. In this paper, we turn to data-driven modeling with neural networks to learn, offline from past data, an adaptive controller with an internal parametric model of these nonlinear features. Our key insight is that we can better prepare the controller for deployment with control-oriented meta-learning of features in closed-loop simulation, rather than regression-oriented meta-learning of features to fit input-output data. Specifically, we meta-learn the adaptive controller with closed-loop tracking simulation as the base-learner and the average tracking error as the meta-objective. With a nonlinear planar rotorcraft subject to wind, we demonstrate that our adaptive controller outperforms other controllers trained with regression-oriented meta-learning when deployed in closed-loop for trajectory tracking control.