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

Discrete time noise covariance #36

Closed andre-nguyen closed 4 years ago

andre-nguyen commented 4 years ago

Hi,

thank you so much for answering questions :smile:. In VINS-Mono the discrete time noise covariance matrix is approximated to the first order using image

So dt * Gt * Qt * Gt' however in your supplementary material your approximation of Qd is image

Where does the extra dt come from?

hyye commented 4 years ago

This is from another convention of discretization. I discussed with the author of VINS-Mono. The first one is more proper. I will change the supplementary material accordingly.

andre-nguyen commented 4 years ago

Interesting, do you know the reference? I think you can leave it like that in your supplementary material because that is how you did it in the code.

hyye commented 4 years ago

For the first one, this paper can be a reference. For the second one, I cannot find a more official paper.

andre-nguyen commented 4 years ago

Thanks!