Weighted Filtering Method for Distributed Cooperative Localization Based on UWB/INS Integration

Ultra-Wideband (UWB) and Inertial Navigation System (INS) integration is widely applied in multi-robot cooperative localization (CL). Reliable information fusion requires an accurate and consistent estimate of the state covariance. However, due to error accumulation during INS propagation, UWB ranging noise, and modeling imperfections, the estimated covariance often becomes inconsistent. To address this issue, this paper proposes covariance-weighted distributed filtering method for multi-robot CL based on UWB/INS integration. First, an equal-weight fusion strategy is employed to conservatively and consistently fuse the state estimates of all nodes. Furthermore, to better exploit covariance information and improve fusion effectiveness, a trace-based weighting strategy is introduced, where the fusion weight is determined by the trace of each node's covariance matrix, yielding more consistent fused covariance and significantly improved localization accuracy. By jointly leveraging INS propagation and UWB pairwise ranging, the proposed framework enables fully distributed CL. Simulation results show that the equal-weight strategy reduces localization error by approximately 2 5 boldsymbol% compared with standalone INS, while the trace-based strategy achieves up to an 80% reduction, significantly outperforming the equal-weight strategy and demonstrating the effectiveness and robustness of the proposed approach.