I have run the foundation pose rosbag demo and rtdetr realsense demo successfully.
However, when I tried to detect my object, no mask was shown as the picture.
I modified isaac_ros_realsense/config/realsense_mono_depth.yaml by adding initial:=true, and filter:=pointcloud to try to solve the control_transfer error. Unfortunately, none of this work.
The camera works fine in realsense-viewer, and the foundation pose can detect the scissors properly with the command I used (/wo making a scissors mesh). I doubt it was the camera’s problem that I can’t detect my own object.
I found this post and modified the isaac_ros_realsense_mono_rect_depth_core.launch.py. Here is the log: log.txt (47.1 KB). Hope this could help figuring out what’s wrong.
Hello, I’m having the a similar problem. when I looked through the launch files, it seems that some are wrongly configure with regard to remapping. I had to add to the realsense_mono_depth.yaml → depth_module.depth_profile: 640x480x15 And also it looks like the launch files wrong for the remapping:
There are 4 inputs required by foundation pose with exact matched timestamps. I suspect one of these, mostly segmentation has some issue. Can you echo a few messages from each of these and post it here?