This paper presents a novel safety control framework for Loitering Aerial Vehicles (LAVs) operating under non-Gaussian measurement uncertainty. The approach integrates variational inference-based belief dynamics with adaptive buffered half-space constraints, transforming complex probabilistic collision avoidance into tractable convex geometric conditions. This ensures rigorous safety guarantees while avoiding the conservatism of robust methods. An event-triggered hierarchical planner further balances global optimality with local responsiveness, enabling rapid navigation in dynamic environments. Validated through 1000 Monte Carlo simulations, the framework achieves a 95.4% success rate. Comparative analysis demonstrates that the proposed method compares favorably with state-of-the-art safety-set approaches by effectively resolving local infeasibility issues and maintaining real-time efficiency without compromising probabilistic safety assurance.
Zhao et al. (Fri,) studied this question.