Multi-Drone Cooperative Localization via UWB and VIO Fusion

Drone localization holds paramount importance in the realm of drone applications. Presently, the prevailing method for single localization relies on visual-inertial odometry (VIO), achieved through the fusion of visual and inertial sensors, which is marred by substantial accumulated errors resulting from the absence of closed loop information. To address the issue of accumulated errors in single drone localization, one well-established approach involves fusing Global Positioning System (GPS) information with visual data. However, in intricate environments like indoors or mountainous areas, GPS fails to provide accurate position information. we propose harnessing the power of cooperative localization involving multi-drone. Consequently, the integration of other sensors becomes imperative to minimize cumulative errors. In this paper, we investigate the utilization of inter-drone distance information as constraints to mitigate cumulative errors in single drone localization. We present an algorithm that merges ultra-wideband (UWB) ranging information with VIO, rectifying the outcomes of single drone localization. In contrast to traditional VIO algorithms, our proposed algorithm demonstrates reduced cumulative errors and substantial enhancements in localization accuracy. Experimental results validate the feasibility of our collaborative algorithm, revealing its superior localization accuracy when compared to the state-of-the-art VIO algorithm, VINS.