[ros-users] [Discourse.ros.org] [ROS-Industrial] Integration of 6-axis manipulator and 2-axis positioner (external axes)
Thomas Lehmann via Discourse.ros.org
ros.discourse at gmail.com
Sun Mar 3 21:21:25 UTC 2019
Thank you very much for the PR, @gavanderhoorn. I've changed `moveit_planning_execution.launch` and related scripts to launch two namespaced instances of `industrial_robot_simulator` when argument `sim` is set to `true`. I had tried to "push down" the simulator and `joint_trajectory_action` server instances individually using `ROS_NAMESPACE` before without success. Your approach using groups seems to work without issues.
I can now successfully plan and execute pose goals for the manipulator and positioner individually in rviz. This should solve the initial issue I described for simulation. Once our system is ready to use, I will carry out the same tests on the physical robot.
> Note: this only gets you to the *actual* problem / challenge: planning for the combined scene (or at least: with kinematics involved).
The end goal for our application of metal additive manufacturing (and I am assuming for many others using the manipulator/positioner combination for various purposes) would be to plan paths not only based on getting from point A to B whilst avoiding collision but planning manipulator and positioner (and thus workpiece) pose trajectories from tool paths that are generated from CAD-2-path tools. Such tools (e.g., slicers such as the [ros_additive_manufacturing](http://wiki.ros.org/ros_additive_manufacturing/kinetic) package) generate tool paths with many via points based on the workpiece geometry. This response by @gavanderhoorn to a post on ROS Answers explains the difference and challenges involved very well IMO: [Planning complex plans using multiple waypoints with MoveIt](https://answers.ros.org/question/261368/planning-complex-plans-using-multiple-waypoints-with-moveit/). Based on that post and my preliminary testing when only considering the manipulator for planning, [De
scartes](http://wiki.ros.org/descartes) seems to be a suitable alternative but also MoveIt! has some capabilities for planning Cartesian paths with waypoints: [`move_group` tutorial: cartesian paths](http://docs.ros.org/indigo/api/moveit_tutorials/html/doc/pr2_tutorials/planning/src/doc/move_group_interface_tutorial.html#cartesian-paths).
I think the challenge as far as I can tell now is the combined planning of workpiece and manipulator trajectories with collision avoidance and then also the synchronization of end-effector commands (e.g., welding torch on/off).
Any ideas for avenues to go down or suggestions of tools that can be used or adapted to ultimately achieve this final objective would be much appreciated.
----
----
> Could you please in case of parallel requests for support through different channels always present/include pertinent information in posts like yours?
For sure, will do that in the future. This particular reply form Eric was part of an email exchange with me a few months ago where I tried to get a sense of the general Yaskawa system capabilities related to ROS.
---
[Visit Topic](https://discourse.ros.org/t/integration-of-6-axis-manipulator-and-2-axis-positioner-external-axes/7931/12) or reply to this email to respond.
More information about the ros-users
mailing list