Why am I getting thick obstacles on RViz?

[stevemartinov]

I am using ZED M camera and don't get me wrong, I think 2D map that I am producing with RTABmap is usable as it clears obstacles but I think I can do much better and have a look at my produced map:


Here is by .db:
https://drive.google.com/file/d/10EORsgbuq13CqInyRxPoFISMc6qItBC8/view?usp=sharing

I have used the regular parameters from rtabmap.launch file and I have also subscribed to odom frame instead of topic as subscribing to the topic breaks my TF>

[matlabbe]

Hi,

The TF of the camera is wrong, it seems the camera is looking done, look how the ground is not aligned with the ground plane:


The reason why the walls look thick is because the ZED depth estimation interpolate a lot in areas in image where there is no texture. Look how the wall on the left and the ceiling are deformed:


You may give a try with rtabmap subscribing to left and right images instead (stereo mode), so that depth estimation is done with cv::StereoBM instead, which doesn't interpolate textureless areas in the image (only edges). There could be also some parameters tuning that could be done on ZED camera driver to make it interpolate less the textureless areas.

cheers,
Mathieu

[stevemartinov]

I changed some parameters and seems to be doing better now.

I have recorded a new db and it seems to be that the cloud is now noiser and the ground is somewhere at the middle.

What could be wrong now? I am using remote mapping and rtabmap uses rgbd_image topic

Me db:
https://drive.google.com/file/d/1d_rHK5s2Rsq9oAZ7FVHkEflhuBXoPZ7Q/view?usp=sharing

[matlabbe]

There are both stereo and rgbd images in the database. How did you launch rtabmap? If you are using "rgbd_image", do you have two nodes publishing on rgbd_image? Also, the camera is still tilted toward the ground. I suggest to subscribe to point cloud of the zed in RVIZ and adjust your TF so that the cloud is aligned with the ground plane. This will help to get an occupancy grid map matching the environment.

Using the stereo mode, the cloud will be more sparse, as only edges would be estimated.

cheers,
Mathieu