5G-Enhanced Indoor UAV Localization and SLAM through Sensor Fusion

Indoor localization and navigation for Unmanned Aerial Vehicles (UAVs) remain challenging due to GPS denial and the limitations of traditional visual-inertial systems. The emergence of 5G networks offers new opportunities for precise indoor positioning, but their integration with existing UAV navigation systems remains unexplored. This thesis systematically investigates the feasibility of integrating 5G Time-of-Arrival (TOA) measurements with advanced sensor fusion techniques to enhance indoor localization and SLAM (Simultaneous Localization and Mapping). Two approaches are proposed for localization: a real-time Error State Kalman Filter (ESKF) framework and a Pose Graph Optimization (PGO) method. The study leverages the EuRoC MAV dataset, augmented with simulated 5G TOA measurements, to evaluate system performance across diverse indoor scenarios and 5G base station densities. Using just IMU and TOA measurements as a minimal sensor setup, both proposed methods demonstrate significant improvements in pose estimation accuracy and drift reduction, with the PGO-based approach achieving superior results, reaching accuracies up to 13 cm with five base stations. A unified SLAM framework is developed that incorporates 5G TOA measurements alongside visual-inertial data, providing global localization capabilities and resolving scale ambiguity in monocular configurations. For local SLAM, the system operates effectively even with unknown base station positions and intermittent connectivity patterns, demonstrating an average 4.40% improvement in local accuracy while maintaining reliable scale estimation. Furthermore, TOA integration serves as an effective alternative to loop closure, improving accuracy by up to 29.6% in scenarios where traditional loop closure is unavailable. Comparative analysis with state-of-the-art approaches confirms the robustness of the proposed methods even under relaxed operational constraints. This research bridges the gap between visual-inertial and 5G radio-frequency-based approaches, establishing realistic baselines for understanding the practical impact of 5G technology on robotic localization and navigation in complex indoor environments.