What this is, and what it is not

It is a twin of the cell: where the arm stands, what it can reach from there, which posture covers a whole job, and where the fence goes. It answers layout questions on the bench, before anything is bolted down.

It is not a robot programmer and it never touches a machine. No KRL, no jogging, no network. KUKA|prc does the programming; the pendant does the jogging.

What is trustworthy here

• The kinematics — forward and inverse, all eight branches, joint limits, posture naming and singularity margins. 58 checks in test/, including an FK↔IK round trip over 400 random poses.
• The published numbers — 100 kg payload, 3500 mm reach, 1245 kg, ±0.06 mm.

What is not, yet

• Link geometry is estimated to fit the published reach. Real dimensions come off the KUKA drawing or a tape measure on the arm.
• Axis limits are the KR 100-2 HA's, a sibling model, not confirmed for this P variant.
• No collision model. Folded-back poses near the base are reported reachable; in reality the arm would hit itself or the pedestal.
• The KUKA A1–A6 sign mapping is unverified. Every angle on screen and in the CSV is shown in KUKA convention, but the A4–A6 signs are a reasoned guess until PRC confirms them.
• The body is a schematic, not the real castings: correct joint frames, approximate shapes. Load the real CAD per link in Robot model — it changes what you see, never the maths.

How it gets verified

Export waypoints, run the same points through KUKA|prc, compare. PRC wins every disagreement. Then the estimated numbers become measured ones and the flags in the header go away.

Close