UWB-Enhanced BundleFusion System for Improvements and Applications in Indoor Environments

Driven by the increasing demand for medical, educational, and public services, household service robots have experienced rapid development. High-precision indoor positioning and real-time 3D reconstruction technologies are essential for enabling intelligent services in these robots. However, despite being regarded as one of the most efficient open-source frameworks for real-time 3D reconstruction, BundleFusion still faces significant limitations in complex indoor scenarios, such as tracking failures, accumulated drift, and insufficient scene detail, which severely restrict its practical applications. To address these challenges, this paper proposes an improved BundleFusion approach enhanced by UWB (Ultra-Wideband) real-time positioning information. By designing a UWB+RGB-D data acquisition system, UWB absolute positioning data is integrated into key BundleFusion modules, including feature matching, hierarchical optimization, dynamic model updates, and loop closure detection. This integration significantly enhances the system's tracking robustness and global consistency. Experimental results demonstrate that the proposed approach effectively resolves tracking failures in challenging scenarios such as sparse features, dynamic object interference, and rapid camera motion, achieving high-precision 3D environment reconstruction. This study provides a novel technical solution for the application of household service robots in complex indoor environments and contributes to the advancement of real-time 3D mapping and high-precision positioning technologies.