Consistent-Innovation-Aided Distributed Cooperative Localization for Multi-UAV Navigation in GNSS-Denied Environments

Cooperative localization provides a promising solution for multi-UAV navigation in GNSS-denied environments. However, centralized cooperative localization and exact cross-covariance-based distributed methods usually require global state management or explicit propagation of inter-node cross-covariance, resulting in heavy communication and computational burdens. To address this problem, this paper proposes a consistent-innovation-aided distributed cooperative localization method for multi-glider UAV swarms. The method uses a strapdown inertial navigation system as the reference source and introduces inter-node ultra-wideband ranging as cooperative constraints. Each node maintains only its local state and covariance, while a conservative innovation covariance approximation is constructed to perform distributed measurement updates without explicitly propagating global cross-covariance. The cooperative localization sub-filter is further integrated with altitude, geomagnetic, and scene-matching sub-filters through a federated filtering framework with adaptive vector information allocation. A GNSS-denied multi-glider release simulation is established to compare the proposed method with non-cooperative localization, centralized cooperative localization, and an exact-cross-covariance distributed baseline. The results show that the proposed method reduces the position, velocity, and yaw RMSEs by 27.14%, 21.90%, and 9.68%, respectively, compared with the non-cooperative method. Compared with the exact-cross-covariance baseline, the proposed method achieves comparable localization accuracy while reducing communication traffic by approximately 91.44%. Packet-loss experiments further show that the proposed method maintains bounded errors under a 20% packet loss rate, demonstrating improved robustness and engineering feasibility for communication-constrained UAV swarm localization.