UAV swarm tracking in low-altitude economies: A robust online learning-based control strategy

With the rapid development of low-altitude applications, ensuring robust control of UAV swarms under uncertain communication environments has become a critical challenge. This paper investigates online robust tracking control for a UAV swarm with one leader and multiple followers communicating over unreliable multiple-input multiple-output (MIMO) channels. A dynamic model is constructed to capture uncertainties arising from both inter-UAV coordination and wireless communication. The control problem is reformulated as a virtual ergodic optimal control task for an auxiliary system that explicitly accounts for these uncertainties, and the optimal solution is shown to guarantee robust stabilization. To address the curse of dimensionality, we derive a reduced-order formulation and develop an online learning framework based on stochastic approximation (SA) with almost sure convergence guarantees. Simulation results verify that the proposed method significantly enhances tracking stability, convergence speed, and computational efficiency compared with widely used baselines under stochastic channel conditions and coupled system–communication uncertainties. This work not only establishes theoretical foundations for robust UAV swarm control but also provides practical strategies for large-scale deployment.