Visual coordinate deviation track

[hjl66662]

Hello, I am using my own data to reproduce lvisam and I used your code. I encountered a problem in the process of running, that is, during the running process, the coordinates of vins will deviate from the track, but it seems to have no impact on the odometry and mapping. The odometry and mapping are still normal, and the VIO part has not reported an error or warning. Could you give me some advice, thank you very much

[epicjung]

Since the algorithm is not tightly coupled. vio and lio can be different. you need to choose the better one

[hjl66662]

Since the algorithm is not tightly coupled. vio and lio can be different. you need to choose the better one

Thanks for your reply. So, is this normal, or is it caused by inaccurate vio odometry. At the same time, why the final result not be affected when the odometry of the vio is inaccurate? If the vio odometer does not restart, it will participate in the scantomap process.