Hello everyone,
I am trying to implement the following simulation. I have been working my way into Ros, Linux and C++ the last few weeks, and want to get going, but still have some difficulties of how to best implement what I want.
The szenario is this: I get the pose and orientation of a small object which is detected by an AI (I do not code this part myself). Once I have received this information I want my UR5 to get the object pick it and place it (it will be sticked onto another object).
So far I got my UR5 running (used moveit) so I can plan and execute things in Rviz and they will be also executed in Gazebo. I think I need a Gazebo simulation. I still don't quite get the difference between Rviz and Gazebo, but as far as I understood, Gazebo is the proper environment for simulation since it respects e.g. physical restrictions, whereas Rviz doesn't...
So right now I can only move the UR5 by using plan and execute in Rviz and then the robot also moves in Gazebo. Do I actually need the Rviz window open or is there a way how I can just do things in gazebo only?
I also tried to load my stl object which I want to pick and place into the scene, which worked fine in Rviz by importing file in the motionplanning section, but it didn't appear in the gazebo simulation. Seeing this makes me think even more: I don't really understand the relation between my Rviz and Gazebo simulators, obviously the robot movements are copied, but the environment isn't...
I guess I stop writing here and hope you could get a rough idea of what I am trying to do and would appreciate any help on this! Thank You!
Best regards
Joel
Edit:
↧