Yes. Both documents show machine to robot, but one describes accepting a request and the other says the area is available. Same arrow, different event.

@ElliotAbbott0074 Make separate rows for request received and resulting state confirmed, with an owner for each. That should clear up the normal handover on paper.

Clearer names won't establish who actually holds the part. Your normal handover could still look complete with custody falling between those rows.

@DavidBennett0780 You're right; I overstated what those rows settle. Add known part location and the point where custody becomes uncertain. Signal definitions are only the start.

There's a single transfer box in our sequence, with no part-location detail. Both documents leave out who resolves custody if that step is interrupted.

On another tending project, we moved a paper part token between boxes. It felt silly until both teams put it in different places.

@AaronAllen0262 Does the token need a box for unknown, then?

@GraceBennett0755 Yes. Unknown is a legitimate state to document. Give it an authorised recovery owner rather than drawing an arrow straight back into normal operation.

Would you put recovery ownership in this table or a separate procedure? I'm trying to keep it readable without hiding the awkward bit again.

@ElliotAbbott0074 Keep the owner visible in the table, beside a procedure reference. Separate detailed instructions are fine when their document link and revision are clear.

And don't let either supplier write 'other system' as the owner. That's the same disagreement wearing a slightly nicer shirt.

@DavidBennett0780 That's close to our current wording, unfortunately. One note says machine recovery, while the machine document points back to the integrator for interrupted transfers.

Put the conflicting passages next to each other in the common draft. Then the suppliers can resolve the wording instead of you interpreting their intentions.

@NaomiChan1115 If communication returns, why wouldn't the sequence simply retry?

A working connection doesn't tell you whether the part changed hands. Retrying needs an agreed state and recovery policy that accounts for the physical transfer.

I've got the review structure I needed: explicit signal meanings with responsible owners in a shared table. That closes my documentation question; the actual interface still needs agreement and validation.

That answers the documentation question without implying the interface is validated. The shared table gives the remaining agreement a clear home.