Radio-based sensing offers compelling advantages over optical systems, particularly in challenging environments where visual solutions face limitations. However, their deployment in real applications was limited by effects such as non-line-of-sight (NLOS) propagation. This dissertation investigates the integration of Ultra-Wideband (UWB) range sensing with inertial measurements for robust localization and navigation. Despite UWB's one-dimensional measurement constraint compared to rich visual data, we demonstrate that strategic fusion with inertial sensors can robustly achieve centimeter-level accuracy. The thesis presents three significant new contributions: (1) a comprehensive framework for indoor localization using low-power UWB-IMU fusion; (2) a novel motion capture system achieving competitive accuracy to visual systems for AR/VR applications; and (3) self-calibrated localization methods for UWB-IMU sensor networks in unknown environments. Through theoretical analysis and experimental validation, we provide solutions to fundamental challenges in sensor calibration, optimal data fusion, and real-time implementation, establishing a foundation for radio-based localization in diverse applications where reliability and precision are paramount.