search

hyye / lio-mapping

Implementation of Tightly Coupled 3D Lidar Inertial Odometry and Mapping (LIO-mapping)
https://sites.google.com/view/lio-mapping
GNU General Public License v3.0
898 stars 320 forks source link

Confused with constraints part #37

Closed getupgetup closed 4 years ago

getupgetup commented 4 years ago

Hi, I'm new in LIO/VIO. According to your paper, I think the most different part between LIO and LOAM is the constraint/factor construction. I found this in your paper: 1 In my opinion, LOAM simply use IMU data to give a predicted pose and focus on the second part of it. LIO add frame-to-frame pose constraint by IMU pre-integration. Am I right? Threrefore, I'm confused with these constraints. the second part represents point-to-plane residuals and third part represents pose-to-pose residuals. Why can they be "added"? How to determine their covariance? Thank you~

hyye commented 4 years ago

This is the pre-integration term, as shown in supplementary material Eqn. (C.8), not pose-to-pose residuals.

getupgetup commented 4 years ago

This is the pre-integration term, as shown in supplementary material Eqn. (C.8), not pose-to-pose residuals.

Thank you! Is this part mainly in function: Estimator::SolveOptimization()?