Fusion Positioning of Ultra-wideband Single Base Station and Visual Inertial Odometry Based on Arrival of Angle and Time of Flight

摘要

To achieve robust and drift-free localization of robots during long-term autonomous navigation in complex indoor environments characterized by dim lighting and low texture, this paper introduces a vision-aided inertial solution that incorporates ultra-wideband (UWB) technology. The proposed system comprises a randomly placed UWB anchor and a visual inertial navigation robot equipped with a UWB tag. In contrast to previous studies that focus on distance-based UWB cooperative visual inertial odometry, this paper utilizes ultra-wideband technology based on Arrival of Angle (AOA) and Time of Flight (TOF). This approach enables simultaneous acquisition of both the distance and orientation of the UWB tag relative to the UWB anchor. The UWB tag, mounted on the robot, captures the robot's position and distance information with respect to the UWB anchor. After preprocessing, these data are integrated with the localization information derived from visual inertial measurements in the SLAM backend, employing a nonlinear optimization-based estimator. The effectiveness of the proposed system is validated on the EuRoc dataset, demonstrating satisfactory localization performance. Notably, in visually challenging environments where visual inertial measurements are prone to significant drift, the proposed method accurately fulfills the localization task.