Which collaborative robot arm would you recommend? [closed]

asked 2017-05-29 08:17:53 -0500

Max Pfingsthorn gravatar image

Collaborative robot arms are all the hype right now, but it is difficult to see through the marketing and choose one based on actually working features without having tried it. Some comments I have received from friends already, dating back up to two years:

  • Kuka LBR iiwa is hard to integrate with external planners like MoveIt (control rate problems).
  • Universal Robots UR5/10 are hard to control externally, because scripts have to be sent to the pendant.
  • Universal Robots "just" check overcurrent, so it is difficult to do force-sensitive work.
  • Franka Emika looks great, but can't tell yet if it will be actually good.

To me, this sounds like the marketing/glossy paper says one thing, while it looks quite different in practice. While integration protocols like Profibus/Modbus seem to be rather well supported across the board, it is not easy to find out how deep the integration goes and which low-level functions can be controlled via the protocol. Also, manufacturers usually provide a way to execute custom programs on the control hardware, and it is often not clear to what extend custom code can be used and what support is available to communicate externally (e.g. via Ethernet).

I am interested if people have recent experience with commercially available collaborative arms like:

Opinions on other robots (FANUC, ABB, F&P Robotics etc) are very welcome as well, though they seem to be less popular.

I'm especially interested in problems you may have faced, what was easy to do, or advice you found useful in the areas of:

  • Integration (electrical connectivity, protocols, API/SDK, etc)
  • Interface (external controllers, planners and/or world models, joint states output, etc)
  • Human-Robot Collaboration (signaling of collisions and external/joint forces, safety parameters/facilities, etc)

Any other advice would also be appreciated!

PS: To preempt the inevitable question about my specific use case: I do not have or want one beyond human-safe manipulation. The basic idea is to have a good robot to do HRI experiments with. Cost is (somewhat) secondary as well. Ideally, this thread would turn into a collection of opinions, user reviews so to say, about the different arms.

edit retag flag offensive reopen merge delete

Closed for the following reason too subjective or argumentative by Max Pfingsthorn
close date 2017-05-30 03:42:23.042432

Comments

Ideally, this thread would turn into a collection of opinions, user reviews so to say, about the different arms.

if that is really your goal, I would suggest you post this to discourse.ros.org and not here, as this is a (single) Q & (single) A site, not a discussion forum.

gvdhoorn gravatar image gvdhoorn  ( 2017-05-29 12:51:33 -0500 )edit

Also: can you clarify what the link with ROS is, as right not I don't see any mention of anything ROS.

gvdhoorn gravatar image gvdhoorn  ( 2017-05-29 12:52:34 -0500 )edit
1

Right, I'll repost to discourse. Thanks. The link is implicit, of course it should work with ROS, incl. MoveIt, etc.

Max Pfingsthorn gravatar image Max Pfingsthorn  ( 2017-05-30 03:41:45 -0500 )edit

re: link: yes, I figured, just wanted to make sure.

gvdhoorn gravatar image gvdhoorn  ( 2017-05-30 03:46:48 -0500 )edit

Was this ever asked on discourse.ros.org? I'd love to get some feedback, but couldn't it find it there.

mkillpack gravatar image mkillpack  ( 2017-11-09 18:13:00 -0500 )edit