RE: Problems interfacing RTAB-Map and V-Rep

Posted by PierHarper on
URL: http://official-rtab-map-forum.206.s1.nabble.com/Problems-interfacing-RTAB-Map-and-V-Rep-tp1444p1459.html

Hi, thanks for your last reply, it helped me a lot!
Unfortunately i got another problem.
The message's data field streamed by simros_strmcmd_get_vision_sensor_depth_buffer contains 480*640*1=307200 elements while the brand new sensor_msgs/Image i've created contains 480*640*4*1=1228800 elements in its data field.
When i put these 307200 depth elements ,multiplied for 300 in order to consider my kinect's range (3m), in the data field of the new image, i noticed that remains 1228800-307200=921600 empty elements.Obviously, if i left empty these 921600 elements and try to run rtabmap and visualize on rviz i got nothing in the sense that there is no 3D map reconstruction. I also got same result if i firstly build an image in 8UC1 format in which i put the depth_values (with no one left empty) and then convert it in 16UC1 format and then in 32FC1 format using cv_bridge (the 3D map is not build and i get some warnings like "util3d_registration.cpp:184::transformFromXYZCorrespondences() RANSAC refineModel: Refinement failed: got an empty set of inliers!" , "[pcl::VoxelGrid::applyFilter] Leaf size is too small for the input dataset. Integer indices would overflow." and errors:"[pcl::IterativeClosestPoint::computeTransformation] Not enough correspondences found. Relax your threshold parameters.") .
How can i solve it?