Author Topic: Camera mounted robot methodology  (Read 3010 times)

AiSard

  • Forum User
  • *
  • Posts: 16
    • View Profile
Camera mounted robot methodology
« on: November 07, 2016, 06:23:42 PM »
Was wondering if there is a way to use kukaPRC to move a camera mounted on a robot in such a way that it is always focused upon a fixed (or moving) point?

I recall doing some tests way back when SPLine was not yet fully implemented in kukaPRC, but gave up as I couldn't find a way to keep the camera focused on a point in between commands. LIN and PTP always shifted the focus in between commands.

Not going to have access to the robot til Thursday so was wondering if it's possible to do on PRC, or if not if you have any tips for how to implement it with KRL code?..

Thanks,

 

Johannes @ Robots in Architecture

  • Lead Developer
  • Administrator
  • Forum User
  • *****
  • Posts: 937
    • View Profile
    • Robots in Architecture
Re: Camera mounted robot methodology
« Reply #1 on: November 07, 2016, 07:05:26 PM »
Hello,

Basically that shouldn't be a problem - you calibrate your tool so that the optical axis is the tool axis, and then you create your toolframes with the Plane-Normal component to have each plane face your object.
I don't quite see what the advantage of a SPLINE movement would be in that regard, though? And you would most likely have to use rather dense toolpaths if you want your camera to follow closely. Alternatively, you can also try setting the TCP to your focal point (i.e. a tool that the robot thinks is 2000mm long or so), that may reduce judder as well.
I would guess that calibration is mostly the issue, and not so much the movement itself.

Best,
Johannes

AiSard

  • Forum User
  • *
  • Posts: 16
    • View Profile
Re: Camera mounted robot methodology
« Reply #2 on: November 08, 2016, 10:26:15 AM »
Hmm.. I've vague recollections that in the preliminary tests, in between the planes (in between the commands) the focus would shift to the left/right before refocusing on the next plane/command. Will have to test further.. hopefully will get enough hands-on time with the robot before actual implementation!

Johannes @ Robots in Architecture

  • Lead Developer
  • Administrator
  • Forum User
  • *****
  • Posts: 937
    • View Profile
    • Robots in Architecture
Re: Camera mounted robot methodology
« Reply #3 on: November 08, 2016, 05:21:08 PM »
Hello,

I've thought about it a bit longer and I'm quite confident that you could solve that by having the TCP at the focal distance of your camera. That's easy if the distance stays the same, but will be more tricky if you want to change it dynamically. Maybe you could use different tools for different distances and just use the Change Tool component to switch them.

Best,
Johannes

AiSard

  • Forum User
  • *
  • Posts: 16
    • View Profile
Re: Camera mounted robot methodology
« Reply #4 on: November 08, 2016, 05:56:46 PM »
mm.. was really hoping there was an easy way for changing it dynamically. Will test out the Change Tool technique and see if it fits the scenario on Thursday.

Thanks,

Johannes @ Robots in Architecture

  • Lead Developer
  • Administrator
  • Forum User
  • *****
  • Posts: 937
    • View Profile
    • Robots in Architecture
Re: Camera mounted robot methodology
« Reply #5 on: November 08, 2016, 06:07:19 PM »
If you change it dynamically, the problem is that you would also have to change the programmed position of the tool. So let's assume you've got a 3m long "tool" and then change it to a 2m long "tool" while the programmed position stays more or less the same. The robot will then try to move 1m closer, which is not exactly what you want.
So you have to componsate that in parallel with the programmed position, which may again cause problems.

But it's been a long day, so I may be overthinking it or missing something obvious...