Complementary Meta-Reinforcement Learning for Fault-Adaptive Control

TL;DR

This paper introduces a meta-reinforcement learning method that rapidly adapts control policies for fault-tolerant systems, leveraging prior fault-specific policies to improve response times and efficiency in critical scenarios.

Contribution

It presents a novel meta-learning approach using a library of prior fault policies, enhancing quick adaptation over traditional MAML methods in fault-tolerant control systems.

Findings

Improved sample efficiency in fault adaptation

Successful application to aircraft fuel transfer system

Faster policy adaptation compared to baseline methods

Abstract

Faults are endemic to all systems. Adaptive fault-tolerant control maintains degraded performance when faults occur as opposed to unsafe conditions or catastrophic events. In systems with abrupt faults and strict time constraints, it is imperative for control to adapt quickly to system changes to maintain system operations. We present a meta-reinforcement learning approach that quickly adapts its control policy to changing conditions. The approach builds upon model-agnostic meta learning (MAML). The controller maintains a complement of prior policies learned under system faults. This "library" is evaluated on a system after a new fault to initialize the new policy. This contrasts with MAML, where the controller derives intermediate policies anew, sampled from a distribution of similar systems, to initialize a new policy. Our approach improves sample efficiency of the reinforcement learning process. We evaluate our approach on an aircraft fuel transfer system under abrupt faults.