GO-SLAM: GPS-Aided Visual-Inertial SLAM for Adaptive UAV Navigation in Outdoor-Indoor Environments

초록

This article introduces a GPS-aided visual-inertial simultaneous localization and mapping (SLAM) system named GO-SLAM. This proposed system optimizes uncrewed aerial vehicles (UAVs)-based search and rescue (SAR) operations by enabling efficient navigation in transition areas between indoor and outdoor environments. Therefore, the UAV addresses challenges posed by intermittent GPS signals and limited features for visual odometry (VO). GO-SLAM optimizes local pose estimation by leveraging available GPS signals and inertial data from visual-inertial odometry (VIO). The system incorporates kinodynamic path search, which employs a hierarchical, time-optimal trajectory planning method to ensure comprehensive coverage. The efficacy of GO-SLAM is evaluated through scenarios involving transition areas with unreliable or absent GPS signals, followed by simulations and real-world experiments to validate its adaptive navigation capabilities.

키워드