Lyapunov Function-guided Reinforcement Learning for Flight Control

2025-10-26

AI Summary

This paper addresses the stability guarantee problem for online flight control systems within a reinforcement learning framework. To mitigate the destabilizing effects of discretization-induced and incremental-model prediction errors, we explicitly model the Lyapunov function’s increment as a convergence metric and embed it directly into the policy optimization objective, enabling joint learning of stability constraints and performance optimization. The proposed cascaded online control architecture ensures both theoretically verifiable Lyapunov stability and action smoothness. Simulation results demonstrate that the method achieves rapid convergence while significantly enhancing closed-loop robustness and control accuracy. It thus provides a practical, data-driven flight control solution with rigorous Lyapunov stability guarantees.

Technology Category

Search and Optimization: Learning to SearchMachine Learning: Online Learning & BanditsConstraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingEconomics, Online Markets and Human Computation: LLM based quality controls for crowd workResponsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

A cascaded online learning flight control system has been developed and enhanced with respect to action smoothness. In this paper, we investigate the convergence performance of the control system, characterized by the increment of a Lyapunov function candidate. The derivation of this metric accounts for discretization errors and state prediction errors introduced by the incremental model. Comparative results are presented through flight control simulations.

Problem

• Enhancing flight control system convergence using Lyapunov functions
• Addressing discretization and state prediction errors in control
• Improving action smoothness in reinforcement learning flight control

Innovation

• Lyapunov function-guided reinforcement learning for control
• Cascaded online learning system with action smoothness
• Convergence analysis accounting for discretization and prediction errors