RTLIO: Real-Time LiDAR-Inertial Odometry and Mapping for UAVs.

Area of Science:

• Robotics
• Sensor Fusion
• Navigation Systems

Background:

• Unmanned Aerial Vehicles (UAVs) typically rely on GPS for outdoor localization.
• GPS-denied environments necessitate alternative localization methods for UAVs to perform feedback control and navigation.
• Existing LiDAR-based indoor localization often lacks the high sampling rate required for UAV feedback control.

Purpose of the Study:

• To develop a real-time LiDAR inertial odometer system (RTLIO) for high-precision, high-frequency odometry.
• To enable robust UAV feedback control and navigation in indoor, GPS-denied environments.
• To improve upon traditional LiDAR-inertial odometry (LIO) systems.

Main Methods:

• Fusion of LiDAR and Inertial Measurement Unit (IMU) sensor data to generate high-frequency odometry.
• Solving cost functions incorporating LiDAR and IMU residuals for accurate pose estimation.
• Implementation of stationary initialization, loop closure, and pose-graph optimization techniques.
• Real-time system development for computational efficiency.

Main Results:

• The RTLIO system successfully generates high-precision and high-frequency odometry.
• RTLIO demonstrated superior performance compared to traditional LIO, exhibiting reduced drift over long-range trajectories.
• Experiments on the KITTI odometry benchmark dataset showed RTLIO outperformed ALOAM and LOAM in terms of lower time delay and enhanced position accuracy.

Conclusions:

• The developed RTLIO system provides an effective solution for UAV localization in GPS-denied indoor environments.
• RTLIO offers significant advantages over existing LIO methods, particularly in achieving high-frequency, accurate odometry.
• The system's ability to initialize while stationary and incorporate loop closure enhances its practical applicability for UAV navigation.