Multi-Sensor Integration for Precision Navigation in GNSS-Challenging Environments

In this dissertation, several integrated navigation systems were developed for precision navigation in GNSS-challenged outdoor environments. The first developed navigation system uses an extended Kalman filter (EKF) to perform a loosely coupled (LC) integration of the INS and LiDAR simultaneous mapping and localization (SLAM). Using the raw KITTI dataset, the integrated navigation system was evaluated for three different driving scenarios that mimic driving in different environments. The developed INS/LiDAR SLAM integrated system outperformed the INS with an average reduction in root-mean-square error (RMSE) in the horizontal and up directions of 88% and 32%, respectively. Subsequently, a dataset was collected using an unmanned aerial system (UAS) equipped with a GNSS/IMU board, a low-cost mechanical LiDAR sensor (Velodyne Puck), and an RGB camera. However, due to a malfunctioning GNSS antenna, the GNSS/INS processing resulted in positioning errors that surpassed 25 km. To recover the precise trajectory of the UAV, a GNSS/INS/LiDAR data fusion algorithm was developed. Firstly, the optimized LOAM SLAM algorithm was used to analyze the LiDAR data. The camera images were then processed using Pix4D Mapper software in the presence of ground control points (GCPs). Two scenarios were considered, namely total GNSS outage and support from GNSS PPP solution. In the second scenario, the accuracy improvement in terms of RMSE reductions was around 51% and 78% in the horizontal and vertical directions, respectively. An enhanced integrated navigation system that fuses IMU, LiDAR, and monocular camera measurements using an EKF was developed. The INS/monocular visual SLAM system used LiDAR depth data to solve the scale ambiguity of monocular visual odometry. The system was tested using two datasets, namely the KITTI and the Leddar PixSet datasets. The system led to a reduction in the INS RMSE by around 80% and 92% in the horizontal and upward directions, respectively. Finally, a robust integrated INS/LiDAR/Stereo SLAM navigation system was developed and tested using the raw KITTI dataset. The system produced better position estimations than utilizing the INS solely. The accuracy improvement was quantified by a reduction in the RMSE by 83% and 82% in the horizontal and up directions, respectively.