search

UditSinghParihar / g2o_tutorial

Tutorial on pose graph optimization using g2o
84 stars 9 forks source link

info_mat #3

Open YJonmo opened 1 year ago

YJonmo commented 1 year ago

Hi,

Thanks for this helpful tutorial. I could follow the edge in terms of the delta_x, delta_y, and delta_theta. But I don't understand what the hard coded info_mat = "500.0 0.0 0.0 500.0 0.0 500.0" is.

Could you clarify why you chose those numbers for the info_mat and what does it do in the optimization?

Cheers, Yaqub

UditSinghParihar commented 1 year ago

Hi, Please look into this issue, there I have mentioned that the absolute values of information matrix doesn't matter, but what matters is the ratio of the values of the information matrix between loop closure and odometry edges. These 500.0 value of information matrix is used from ORB SLAM and RTABMAP code. You can change these values accordingly, but ensure that you trust your loop closures edges more compared to odometry edges, hence loop closure edges should have higher information matrix value.

Cheers, Udit