Start the proposed interface at one representative inspection station. Link states and decisions to equipment, work orders and workpieces. Check clocks, inspection recipes and interface versions. Define fields and update rules for states, results and evidence records. Site integration and full-line load still need testing.

Define how states and results work together

Record ready, busy, fault and recovery states. Record the decision, validity, result identifier and part identifier. The receiver checks whether the state has expired and whether the result belongs to the current part and inspection cycle. Use a separate status for missing results during faults.

Keep records of receipt and action

Use the result identifier to connect the records of sending, acknowledgment, execution and errors. If the project retains evidence images, index them by result identifier. Define read-only access and retention periods. Record when an image is missing. Base reports on the records actually kept.

Test relevant combinations

Check normal operation, late results, duplicates, network loss, reconnection and restart. Verify that a result for an affected part is not applied to a later part. Keep time, message and action records from both endpoints. Check capacity and recovery at the target cycle and workload.

Related service: Vision algorithms and model engineering.