Using ABB's Absolute Accuracy (603-1) Within a URDF

[jbeck28]

ABB sells a feature, Absolute Accuracy, which involves some calibration routine they perform. I've found what I believe are the values that came with my robot (which has AbsAcc) inside of a MOC.cfg that appears inside of backups. Here's the relevant data:

PARALLEL_ARM_CALIB:
 
       -name "rob1_2" -error_length 0.000251949
 
       -name "rob1_3" -error_length 5.31822E-05
 
 ARM_CHECK_POINT:
 
       -name "rob1_3" -position_x -0.217 -position_y 0 -position_z 0.2154
 
 ARM_LOAD:
 
       -name "r1_load_1"
 
       -name "r1_load_2"
 
       -name "r1_load_3"
 
       -name "r1_load_4"
 
 ARM_CALIB:
 
       -name "rob1_1" -error_offset_x 2.76348E-05 -error_offset_y 9.85993E-05\
       -error_offset_z -0.000325266 -error_roll 8.72513E-05\
       -error_pitch 0.00078713 -error_jaw 0.000384148
 
       -name "rob1_2" -error_offset_x -0.000104688 -error_roll -8.16564E-05\
       -error_pitch -0.000354871
 
       -name "rob1_3" -error_offset_z 1.49512E-06 -error_roll 0.000220894\
       -error_pitch 0.00308007 -error_jaw 0.000179515
 
       -name "rob1_4" -error_offset_y -0.000165182 -error_offset_z 0.00011376\
       -error_roll -0.000466978 -error_jaw -0.000281756
 
       -name "rob1_5" -error_offset_x -0.000169532 -error_offset_z 0.00021558\
       -error_pitch 2.60882E-05 -error_jaw 6.11623E-05
 
       -name "rob1_6" -error_offset_y 4.64985E-05 -error_offset_z 5.62064E-05\
       -error_roll -0.00067455 -error_jaw -8.66063E-05
 
 JOINT_CALIB:
 
       -name "rob1_1"
 
       -name "rob1_2" -Compl 2.12623E-06
 
       -name "rob1_3" -Compl 9.58023E-06
 
       -name "rob1_4" -Compl 4.50031E-05
 
       -name "rob1_5" -Compl 5.08124E-05
 
       -name "rob1_6" -Compl 0.000143289

The question, if you will, is how properly to apply these? It seemed at first glance to be as easy as applying these to the joints within the URDF. However, this does not give me a match with EEF locations between robot controller and ROS - they differ by roughly a millimeter. I then stumbled upon this, and realized the roll pitch and yaw might not be what I think they are in the URDF. I manually rearranged RPY to match the frames given in that diagram, and got a very different result that a similar order of magnitude of wrongness.

I've got a ticket open with ABB to see if they're willing/able to help, I'll update this once I hear back either way.

Thanks again,
Josh

[gavanderhoorn]

I've got a ticket open with ABB to see if they're willing/able to help, I'll update this once I hear back either way.

@jbeck28: did ABB get back to you?