A GNSS/INS/LiDAR Integration Scheme for UAV-Based Navigation in GNSS-Challenging Environments

Area of Science:

• Robotics and Autonomous Systems
• Geomatics Engineering
• Aerospace Navigation

Background:

• Accurate pose estimation is critical for unmanned aerial vehicle (UAV) navigation, with Global Navigation Satellite Systems (GNSS) commonly used for outdoor localization.
• Sole reliance on GNSS poses safety risks due to potential receiver malfunctions or antenna errors, necessitating robust alternative or supplementary navigation solutions.
• Unmanned aerial system (UAS) data collection using GNSS/Inertial Navigation System (INS), LiDAR, and high-resolution cameras highlighted challenges during GNSS signal outages.

Purpose of the Study:

• To develop and evaluate a multi-sensor integrated navigation system for UAVs to ensure reliable pose estimation during GNSS signal outages.
• To address significant trajectory errors (exceeding 25 km) encountered during GNSS/INS processing due to antenna malfunction.
• To compare the performance of the integrated system under complete GNSS outage versus scenarios with GNSS Precise Point Positioning (PPP) assistance.

Main Methods:

• Implemented a multi-sensor integration system combining GNSS/INS, LiDAR (Velodyne Puck), and a high-resolution camera (Sony a7R II).
• Processed LiDAR data using the optimized LOAM SLAM algorithm for position and orientation estimation.
• Utilized Pix4D Mapper software for camera image processing with Ground Control Points (GCPs) to establish precise camera poses as ground truth.

Main Results:

• The integrated GNSS/INS/LiDAR navigation system successfully recovered precise UAV trajectories despite prolonged GNSS outages.
• Significant improvements were observed with GNSS PPP assistance compared to complete GNSS outage, including RMSE reductions of ~51% (horizontal) and ~78% (vertical).
• Root Mean Square Error (RMSE) for roll and yaw angles decreased by 13% and 30%, respectively, while pitch angle RMSE increased by ~13%.

Conclusions:

• Multi-sensor integration, particularly incorporating LiDAR SLAM, offers a robust solution for UAV navigation during GNSS signal degradation or failure.
• The developed GNSS/INS/LiDAR system demonstrates enhanced reliability and accuracy for UAV pose estimation in challenging environments.
• While overall accuracy improved, specific angular error characteristics (e.g., pitch) warrant further investigation and refinement in sensor fusion algorithms.