The trace shows my index update occurs before placement is confirmed. That means the retained number can identify an intended next pocket without establishing that the preceding transfer completed.

Map the transfer stages offline, with an agreed completion confirmation. Have your integrator define how saved state is reconciled with tool and pocket conditions; unknown conditions need an explicit recovery route.

@DanielBell0652 So just moving the increment isn't enough? i'd hoped this was one badly placed line, which would've been embarrassing but convenient.

@MiaBrooks0818 Save the stage immediately after every action and you should always know where to resume.

@DanielBell0652 Updating a stage after an action doesn't eliminate the interval between the physical event and persistence. An interruption there can leave the saved stage inconsistent with reality.

You're right; saving stages alone doesn't guarantee recovery. The saved stage guides reconciliation, and the physical conditions still need confirmation before an approved next action.

On my setup, one index could mean either 'still carrying' or 'already placed'. Giving those states names made the design discussion much less circular.

@WillBennett0748 Our test actually stopped with the workpiece held while the display said to start another pickup. The recovery message was confidently wrong.

Can recovery infer a held workpiece from the gripper's closed indication, or does that indication fail to establish actual possession?

@BethAdams0150 The meaning depends on the installed sensing arrangement. You need the integrator to establish what evidence distinguishes a held workpiece from an empty or indeterminate tool, rather than equating closure with possession.

i still can't establish a reliable recovery decision from the saved and physical states. i'm keeping the job out of normal production while that remains open.

Understood. Without a reliable next-step decision, keeping normal production on hold follows from the gap.