FT‐LVIO: Fully Tightly coupled LiDAR‐Visual‐Inertial odometry

Abstract In this paper, we propose a fully tightly‐coupled multi‐sensor fusion framework termed FT‐LVIO, that fuses measurements from light detection and ranging (LiDAR), monocular camera an inertial measurement unit (IMU) simultaneously to achieve robust accurate state estimation in real time. FT‐LVIO is built atop the of error‐state‐iterated Kalman filter. To take full advantage complimentary characteristics individual sensors, LiDAR point clouds are undistorted by IMU prediction nearest exposure time filter updated with all sensors. addition, efficient sampling method for point‐to‐plane proposed, which can help select providing sufficient constraints pose facilitate low‐drift odometry. Extensive experiments performed on both public NTU dataset private handheld dataset, results show proposed outperforms state‐of‐the‐art LiDAR‐inertial, visual‐inertial LiDAR‐visual‐inertial methods accuracy robustness. Furthermore, survive challenging staircase environment.