Experimental Evaluation of Radar-Based Velocity Estimation for Navigation in GNSS-Denied Environments
Garrett Andrew Mongold · VTechWorks (Virginia Tech) · 2026
Resilient navigation in GNSS-denied environments is critical for safe autonomous ground vehicle operation. LiDAR, radar, cameras, and magnetometers fused with an IMU can provide a navigation solution that limits the inertial drift to maintain safety. Among these sensing modalities, radar is a particularly promising technology as it performs well in poor weather conditions and directly provides relative velocity measurements. This thesis develops a multi-sensor data collection platform to enable radar-inertial algorithm development. Datasets are collected in various environments and include raw LiDAR, radar, and IMU data along with corresponding ground truth. Using these datasets, this work evaluates radar-based vehicle velocity estimation for use as a measurement update within an Error-State Kalman Filter (ESKF). An offline lever arm and mounting angle calibration, posed as a multi-variable optimization problem, is introduced to better characterize radar-to-IMU lever arm and mounting angle parameters, improving measurement consistency. Furthermore, a radar-based vehicle yaw-rate estimation method is proposed and validated against ground truth. The results demonstrate the potential of radar measurements to aid in navigation through GNSS-denied conditions.