Complementary Meta-Reinforcement Learning for Fault-Adaptive Control
This addresses fault-adaptive control for safety-critical systems like aircraft, but it is incremental as it builds upon existing MAML methods.
The paper tackles the problem of quickly adapting control policies in systems with abrupt faults and strict time constraints, presenting a meta-reinforcement learning approach that improves sample efficiency by using a library of prior policies, and demonstrates this on an aircraft fuel transfer system.
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.