lianghongzhuo / PointNetGPD

PointNetGPD is an end-to-end grasp evaluation model to address the challenging problem of localizing robot grasp configurations directly from the point cloud.
https://lianghongzhuo.github.io/PointNetGPD/
MIT License
327 stars 73 forks source link

some questions about 'cam_pos' in kinect2grasp.py #58

Closed xmkkkkkk closed 1 year ago

xmkkkkkk commented 1 year ago

Hello Mr. Liang, thank you for your PointNetGPD.I have some doubts on line 406 in kinect2grasp.py about 'cam_pos'.Do you mean I need to use tf to get the position between ‘/table_top’ and ‘/kinect2_ir_optical_frame’ or something else? But tf doesn't seem to work properly in python3. I'm sorry that my similar knowledge understanding is very shallow, if you can, could you please explain me in detail what do i need to change 'cam_pos' into and how to do it?Thanks again, looking forward to your reply

lianghongzhuo commented 1 year ago

You can use ros-noetic, it support python3

xmkkkkkk commented 1 year ago

Hi Mr. Liang,I catkin_make the source code of the tf package with python3, and import tf will not report the error. I have encountered a new problem. I only have the realsense camera, where do I change the points cloud topic of the kinect camera to realsense points cloud topic in kinect2grasp.py?

lianghongzhuo commented 1 year ago

This is the technical question related to the basic use of ROS, I recommend you ask these questions at https://answers.ros.org/questions/ or search online.