Closed majiqiang closed 4 years ago
Hi @majiqiang
Thanks for showing interest on our project. Unfortunately, I cannot provide you with the exact copy of our architecture since there are confidential data regarding our laboratory (i.e. IPs, ports...). Basically, you can use the current geograsp/src/cloud_processor.cpp as the _pointcloudlistener in the paper. Add to it a publisher node that publishes a custom message (_sceneobjects in the paper) which holds the grasp configuration data.
Then, create some other node (_allegro_plangrasp in the paper) that subscribes to that topic with the grasp configuration data and finds the way to translate our calculated grasp pose to your robotic system. This can simply be using the data in the object GraspPose, transform it from the camera's reference frame to your robot or world, and then find a trajectory to move your robotic grasper to this pose (we used MoveIt! for that).
Hi, Brayan, I'm very interested to your project about Geograsp, and I read your that paper completely. Could you provide your complete solution architecture to me? I want to know how to define some ROS nodes such as “pointcloud_listener” you mentioned. Thank you very much!