Fixed-Time Reinforcement Learning Framework for Containment Coordination of Networked Aerial Vehicles in Inspection Tasks

초록

This paper investigates a fixed-time reinforcement learning framework for containment coordination of networked aerial vehicles in dynamic inspection tasks. The proposed framework integrates sliding mode concepts to enhance the robustness of the system. To address the challenge of limited leader information sharing among followers, a distributed estimation mechanism is designed to reconstruct leader states in a fixed time, thereby eliminating the need for fully informed followers. Based on this, a critic-only adaptive dynamic programming (ADP) framework is employed to learn optimal coordination strategies for both containment behavior and attitude tracking under time-varying environments. Within this ADP framework, the critic neural network (NN) weights are adjusted according to a fixed-time convergent update rule constructed from auxiliary errors associated with the Bellman residual. The framework ensures that the networked aerial vehicle system converges within a fixed time, and the corresponding convergence properties are rigorously verified through comprehensive mathematical analysis. Simulation results demonstrate that the proposed method achieves rapid convergence and improved containment performance in inspection tasks.

키워드