Any way to initialize with rgb-d camera's depth?

[ozakiryota]

I use LSD-SLAM with a rgb-d camera RealSense. Is there any way to initialize the depth map with rgb-d sensor's values instead of the random initialization? I would like to improve the trajectory at beginning and get absolute scales, not relative. Thanks.

[bespoke-code]

If you want to use a depth camera to map the environment, you should take some serious look at TUM-Vision's Volumetric 3D Mapping in Real Time paper and repository.

You could potentially initialize the depth map using the depth sensor, but then, what's the point of using a SLAM algorithm for mapping? What's more, it may involve a lot of trial and error to make it work :)

I hope this helps!

[ozakiryota]

Thank you for your reply and help. I’ll take look at it.