Grasping for a simulated robot [closed]
I'm relatively new to ROS. I'm trying to implement basic grasping using a seven DOF arm. By grasping I mean, perception of the environment, segmenting out the object from sensor data (point cloud), and publishing joint velocities and grasping a simple object (a cube, say) using the arm's gripper.
I tried this by trying to emulate what has been done in http://docs.fetchrobotics.com/gazebo.... for my own robot (simulation in gazebo, not real) which is a Pioneer3DX with a cool1000 arm (7 DOF) attached along with a head_camera for perception. The Fetch Robot in the URL uses a package called simple_grasping which works very effectively for that robot but I'm running into several problems when I try to implement it for mine but I have no idea how to start debugging.
simple_grasping uses several actions to achieve grasping, like FindGraspableObjectsAction, FollowJointTrajectoryAction etc. I was wondering if there was anyone who has worked with this (implementation on Python) or any documentation available to help me understand these better.
If not, what is the simplest way to achieve grasping? I've been working on this for quite some time and couldn't seem to find any well documented method for grasping for a custom robot. Can the MoveIt package on its own achieve grasping? If so, is there a comprehensive guide on how to implement the same?
I work with a 6DOF arm and implemented a quite decent control starting with this tutorial: It contains pretty much everything you need to know about moveit. http://docs.ros.org/kinetic/api/movei...
There is a Smart Grasping Sandbox built by Shadow Robotics that maybe can help you on this.
You can see the code, modify, adapt to your own needs, etc.
Closed as the this question is too open-ended (Please use http://discourse.ros.org/ instead). See https://answers.ros.org/question/265116 also the question the same author opened