BASE Frame Angle routine :

    Good morning forum,

    Tech Specs:

    KR10 - R1100 Agilus-2

    KRC4 Compact

    KSS 8.6.6

    KOP:

    SafeOperation 3.5

    RoboTeam 3.1

    ForceTorque Control 4.1

    Description:

    Running into a snag with our cell.

    The part were wanting to dock up-to is wheeled into place and we use vision to locate and offset our BASE_DATA[Calculated] ( using a 2D camera ).

    My "Problem" being, the Staubli tool changers we're using are somewhat a super interference and "accurate" fitment, and what I'm seeing is that different presentation positions / different floor area's do present varying presentation angles which are causing a headache.

    I'm being asked to software fix a mechanical design problem ( well that's what I fell at this point in time ), but in the spirit of teamwork, I'm wanting to negate as much variance as possible.

    I recall in nuclear doing BASE origin and angle calculations in the past using an inductive switch and fast measurement input (X33). Tho I didn't write this function, it worked really well, and thus wanted to ask if anyone has a piece of code I can deploy and amalgmate into my routine?. Essentially, I'm wanting to calculate the A/B/C and assign to BASE_DATA[Calculated].

    Many thanks in advance

    You have a 2D camera (I assume X, Y, A), but your part placement error is in all 6 degrees of freedom? Yeah, that's a design error all right.

    I don't think there's any quick or simple way to do what you're attempting. This is what KUKA sells TouchSense for, and that's a non-trivial package.

    You have Force-Torque Control -- can you use that to let the robot "feel" its way into engaging the part?

    Upgrading to a full 3D-6DOF vision system would be another way to fix the problem.

    If you have to try a pure software approach with a touch probe... well, I assume you'd use the camera first to get the initial Base correction. You would have to touch multiple points, and do a lot of trigonometry to process the delta between the points "real" positions and "ideal" positions into A, B, and C corrections, then apply those to the Base. It should be possible, but it's not simple.

    Hi Skyfire,

    The designers to be fair "assumed" the X/Y/A would differ between operators and how the present, but now that the tooling has come to site, the interference fitment is quite impressive from an engineered perspective ( as I guess you'd hope from a tool changer ) , but far more suited to a CMM machine sat on a granite worktop LOL!!!!....

    I was wondering if there was anything within a .SRC called FrameLib ( maybe called something similar ).

    also the design engineers are looking into compliance tool changer adaptors ( jury is out on this atm ).

    "You have Force-Torque Control -- can you use that to let the robot "feel" its way into engaging the part?"

    I have seen this is a sales promotion of this tech package placing circuit boards onto mtother boards - has anyone used this "sensor guided motion" before?.

    Where did this FRAMELIB.SRC come from, and what does it contain? It doesn't sound like a standard KUKA file.

    There was, IIRC, an undocumented .SRC file (whose name escapes me) that used to come installed on every KRC, that contained a set of routines for creating an altered Base from a set of "ideal" points and a parallel set of "found" points -- this routine would do a 6DOF best-fit and return an altered version of the original Base that had been shifted&rotated to match the "found" points.

    The H70 package for TouchSense did this, but trying to reverse-engineer that code was a good way to go cross-eyed.

    Morning Skyfire,

    I'll dig through some old customer archives I done for KUKA Systems. The nuclear jobs used a "touch sense" approach that effectively found an ORIGIN via 2 point sensing in working direction #X and #Y then used that result to do a Z 3-point top down approach to create the plain relative to the working base.

    Will report back.

Advertising from our partners