Loosely Coupled Localization System Based on Multi-Base Station UWB and Point-Line VIO

Accurate state estimation during the long-term operation of a monocular visual-inertial navigation system (VINS) in complex environments is a challenging task, to achieve globally accurate and drift-free localization, we employ a Multi-Base Station Ultra-Wideband (UWB) positioning system to provide global constraints for the local visual-inertial navigation system. To address the need for precise localization in low-texture and low-light environments, we propose a loosely coupled localization system that combines Multi-Base Station UWB and Point-Line Visual-Inertial Odometry (VIO). This system integrates the global Multi-Base Station UWB positioning system with the local Visual-Inertial Navigation System. Firstly, the front-end VIO fuses line and point features to extract structured feature information in low-texture environments, enabling more accurate local poses in complex environments. Secondly, the Multi-Base Station UWB positioning system provides global poses in the localization scenario. Finally, local VIO poses and global UWB position information are fed into the loosely coupled framework, facilitating global nonlinear optimization to obtain accurate global pose. We compare our system with existing frameworks on challenging public datasets, and in the same environments, our system exhibits lower localization errors.