FREE-LIVO: Physically Decoupled Yet Tightly Coupled LiDAR-Inertial-Visual Odometry
FREE-LIVO: Physically Decoupled Yet Tightly Coupled LiDAR-Inertial-Visual Odometry
This research presents FREE-LIVO, a tightly coupled odometry method that does not require a rigid LiDAR-camera rig or time synchronization. By using a pretrained image feature matcher to establish correspondences between camera images and dense LiDAR intensity images generated from a single scan, the method constructs reprojection factors to jointly optimize LiDAR and camera poses with IMU constraints. Tested on six sequences from the M3DGR dataset, FREE-LIVO achieves an average absolute trajectory error (ATE) RMSE of 0.45 m without extrinsic parameters and maintains accuracy under time synchronization errors up to 0.5 s. Experiments also show the method can jointly estimate trajectories for LiDAR and camera mounted on separate robots.
The paper introduces FREE-LIVO, a LiDAR-inertial-visual odometry method that operates without rigid sensor mounting or time synchronization by using generative models to create point-to-pixel correspondences; it reports an average ATE RMSE of 0.45 m on six M3DGR sequences and demonstrates joint trajectory estimation for independently moving sensors.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
time synchronization errors
≤0.5
View original evidence
al accuracy degradation under time synchronization errors of up to 0.5 s. Furthermore, experiments with the LiDAR and camera mounted on separate robots demonstrate that the method can jointly estimate the trajectories of independently moving sensors.
Open source S5absolute trajectory error RMSE
0.45
View original evidence
different acquisition times. These factors are jointly optimized in a single factor graph together with constraints from point cloud registration, visual feature tracking, and the IMU measurements associated with each sensor. On six sequences from the public M3DGR dataset, FREE-LIVO achieves an average absolute trajectory error (ATE) RMSE of 0.45 m without provided LiDAR-camera extrinsic parameters and exhibits minim
Open source S4
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
different acquisition times. These factors are jointly optimized in a single factor graph together with constraints from point cloud registration, visual feature tracking, and the IMU measurements associated with each sensor. On six sequences from the public M3DGR dataset, FREE-LIVO achieves an average absolute trajectory error (ATE) RMSE of 0.45 m without provided LiDAR-camera extrinsic parameters and exhibits minim
Open source S4
al accuracy degradation under time synchronization errors of up to 0.5 s. Furthermore, experiments with the LiDAR and camera mounted on separate robots demonstrate that the method can jointly estimate the trajectories of independently moving sensors.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org