Re: RTAB-Map and Unity integration

Posted by viniciusbs on
URL: http://official-rtab-map-forum.206.s1.nabble.com/RTAB-Map-and-Unity-integration-tp4925p5016.html

Hi,

Thanks for the tips.

matlabbe wrote

You are using the Asus, it has quite distortions after 3/4 meters. Maybe just align well the camera so that camera cloud and scan match well under 4 meters. When rendering the clouds form the camera, set a maximum depth to to 3/4 meters (to show in 3D map only points from the camera matching well the scans). Depth distortion could be decreased using the CLAMS approach integrated in RTAB-Map: https://github.com/introlab/rtabmap/wiki/Depth-Calibration

I've given an initial shot at the CLAMS calibration, following the link you sent, but no success so far (I'll try more tomorrow).

But in the meantime, I wanted to check if it was possible to limit the points used by rtabmap to a "less-distorted" region, i.e. center of the image and close to the camera, and check if is there any improvement. I can't seem to find out the correct parameters to do that though. I've tried Grid/RangeMax, Grid/DepthRoiRatios, Kp/MaxDepth and Vis/MaxDepth, but I couldn't notice much change. Is there some other that I'm missing?

matlabbe wrote

For "the point clouds seem to don't update", which topic are you referring? The generated cloud /rtabmap/cloud_map should be updated on loop closure.

For example, in your previous reply here, on the second figure, you mentioned how the scans become correctly aligned after the loop closure, but the point clouds remained misaligned. I was wondering if this is just because of the camera distortion/tf problem or is there some parameter that I could change to improve this.

Cheers,
  Vinicius