search

stella-cv / stella_vslam

This is a unofficial fork of OpenVSLAM (https://github.com/xdspacelab/openvslam)
https://stella-cv.rtfd.io/en/latest/
Other
934 stars 398 forks source link

Describes how relocalization can fail if the angle from the candidate keyframe is too large #143

Open ymd-stella opened 3 years ago

ymd-stella commented 3 years ago

In the case of LiDAR SLAM, the most commonly used LiDAR is a horizontal field of view of 270 to 360 degrees. In this case, if we use a map created by going through a road only once, localization is possible when going through the same road in the opposite direction. On the other hand, in Visual SLAM, a camera with a relatively narrow horizontal field of view of 80 to 150 degrees may be used. In this case, if we use a map created by going through a road only once, it is impossible to localize when going through the same road in the opposite direction. Furthermore, based on the same principle, if the field of view is too narrow, "relocalization can fail if the angle from the candidate keyframe is too large."

It would be useful for those who are not familiar with Visual SLAM if these facts are explained using illustrations.

jeffin07 commented 3 years ago

@ymd-stella Can you please explain this issue,I would like to work on it

ymd-stella commented 3 years ago

@jeffin07 Updated the Description.

jeffin07 commented 3 years ago

@ymd-stella Thank you for the description.How are you going to expect the illustrations?. It would be helpful if can you please give me any sample of similar ones

ymd-stella commented 3 years ago

Although there is no specific example, I think it would be easier for the reader to understand if you show a illustration that is similar to the actual drawing result of Pangolin Viewer or Socket Viewer.