Apply robodk-measured arm kinematics to KRC2ed05

Working with a KRC2ed05, with a KR200-3 comp arm.

I used roboDK's calibration program + laser tracker to create a kinematics DH table:



How can I update the machine data to match this?

some qualifiers:

I have a backup HDD.

I can use the robodk filter and avoid this, but I'd prefer to correct the kinematics at the source, if it's possible.

The closest I've gotten is this thread: kinematic parameter stored inside robot, and acceleration and torque signal channel before and after gear reduction but I don't know exactly how the variables are mapped.

It's easy enough to assume my value of "a'" for Joint 3 (45.091445mm) would replace the even 45mm below in my $TX3P3, both negative. Past those few obvious ones, is there a standard or diagram that shows which values are which? I understand D-H kinematics at a fifth grade level, but these two DH values are missing a theta value anyway. I'm hoping I can at least correct the lengths and take out 80% of my position error.

FRAME $TIRORO={X 0.0,Y 0.0,Z 750.0,A 0.0,B 0.0,C 0.0}

FRAME $TFLWP={X 0.0,Y 0.0,Z 210.0,A 0.0,B 0.0,C 0.0}

FRAME $TX3P3={X 1000.0,Y 0.0,Z -45.0,A 0.0,B 90.0,C 0.0}

REAL $LENGTH_A=350.0 ;GRUNDACHSLAENGE A

REAL $LENGTH_B=1050.0 ;GRUNDACHSLAENGE B

DECL DHART $DH_4={DHART_A 0.0,DHART_D 0.0,DHART_ALPHA 90.0}

DECL DHART $DH_5={DHART_A 0.0,DHART_D 0.0,DHART_ALPHA -90.0}

I tried changing a few of the values as mentioned above, after cold booting I get the errors:
1) Failed to initialize the dynamic model (code E5)

2) erroneous correction data

Resetting the changed values takes care of the errors. Not as simple as I'd hoped.

It's nice to be on the right path for once. I have to get some parts machining before I can finish this tonight, I'll look through the ROBCOR file has a complement for the kinematics.

I tried with just the length_A altered and got the same errors. The fallback plan of robodk's filtering is the more common one anyway, I just didn't want a third step between CAM and running the robot. If it's a pain, maybe we'll upgrade to the spatial accuracy add-on computer (I think you had linked to it in another thread, panic mode).

We'll see what Kuka support says.

Just found some corresponding values in ROBCOR:

$DYN_DAT[13]=350.0
$DYN_DAT[14]=1050.0

if I update these along with the matching values in MACHINE.dat (Length A and Length B) then the error disappears.

Same with a few numbers in the other variables of MACHINE, but I don't know yet how the DH values are accounted for.

Quote from MOM

some of your dh parameters are completely off.

Before overwriting the official ones you should try to solve this problem first

You mean the ones in the photo I attached? Those were generated by RoboDK, I haven't figured out how to convert from their 4 parameters to Kuka's three DH parameters.

On that note, I got the machine.dat manual from Kuka today, here's the relevant diagram. I'll dive into this later in the afternoon.

Quote from panic mode

some of the values you have are 2.7m which is way more than any dimension in that diagram

sheesh, I didn't realize what I was looking at there, that's huge.

Quote from SkyeFire

we had a 6DOF weighted-average algorithm that took the location the robot wanted to go to, averaged out the corrections from the nearest N points in the sample set, and applied them.

This is about what RoboDK does I believe, I was hoping to get around having to run all our CAM programs through their offline filter. It's effective, just a third step to take! I'll give the DH parameters a little more thought, but not hold it as an absolute goal.

Thanks for the info guys.