COOPERATIVE NAVIGATION METHOD FOR UAV SWARMS UNDER DIRECTIONAL ANGULAR SPAN CONSTRAINTS

Abstract

In GNSS-denied environments, the performance of cooperative navigation for UAV swarms is significantly constrained by the geometric observation configuration of ranging measurements. Existing factor-graph fusion frameworks generally adopt a full-measurement-edge strategy; however, under configuration degeneration, ranging noise is amplified by the geometric dilution of precision along weakly constrained directions, leading to a sharp degradation in positioning accuracy. Starting from the linearization of the ranging equation, this paper reveals that a single ranging edge provides only a one-dimensional radial constraint, establishes a quantitative relationship between directional distribution and constraint efficacy, and proposes two metrics—the maximum spatial angular span and the minimum pairwise angle—to characterize the configuration quality. Accordingly, a Direction-Distribution Criterion-based dynamic ranging-edge Selection algorithm (DDCS) is designed, which, through three progressive criteria—local redundancy elimination, global coverage maintenance, and capacity upper-bound control—adaptively constructs the optimal ranging-edge set at each time instant. The algorithm is decoupled from and executed serially before factor-graph optimization, ensuring high computational efficiency. Simulation results demonstrate that, in dynamic swarm scenarios, DDCS reduces peak positioning errors by over 60% during configuration-degeneration periods, and improves the overall average accuracy by approximately 55% compared with full-measurement fusion, effectively sup-pressing accuracy deterioration caused by poor configurations. This work provides a robust observation-selection scheme for resource-constrained UAV swarms in cooperative navigation.