Re: cv::Exception Airsim implementation
Posted by matlabbe on
URL: http://official-rtab-map-forum.206.s1.nabble.com/cv-Exception-Airsim-implementation-tp10281p10304.html
I could not reproduce the crash. You built rtabmap from source, if you have more than one opencv installed, that may be the cause of the crash. Try rtabmap_ros binaries.
Otherwise than that, the depth image looks wrong. Here a closer look:
1. The grid I show is 50 meters x 50 meters, so there is clearly a scale issue in depth values
2. The flat surfaces look curvy in the point cloud, but not in the image, maybe a mismatch between the output calibration values and the ones actually used by the simulator to generate the depth image.
3. The camera is looking up, but frame_id is set to base_link. It means there is a missing optical rotation (RPY: -1.57 0 -1.57) between base_link and the frame_id inside the image.
You can debug these kind of stuff just with RVIZ by using DepthCloud display (the grid there is 10x10 meters):
URL: http://official-rtab-map-forum.206.s1.nabble.com/cv-Exception-Airsim-implementation-tp10281p10304.html
I could not reproduce the crash. You built rtabmap from source, if you have more than one opencv installed, that may be the cause of the crash. Try rtabmap_ros binaries.
Otherwise than that, the depth image looks wrong. Here a closer look:
1. The grid I show is 50 meters x 50 meters, so there is clearly a scale issue in depth values
2. The flat surfaces look curvy in the point cloud, but not in the image, maybe a mismatch between the output calibration values and the ones actually used by the simulator to generate the depth image.
3. The camera is looking up, but frame_id is set to base_link. It means there is a missing optical rotation (RPY: -1.57 0 -1.57) between base_link and the frame_id inside the image.
You can debug these kind of stuff just with RVIZ by using DepthCloud display (the grid there is 10x10 meters):
| Free forum by Nabble | Edit this page |