A robot end-effector in an inspection fixture with a contact probe, reference artifact and data-acquisition box.

What you need

Use a requirements table, versioned configuration records and raw test data. Assign a responsible reviewer for safety-related work.

Illustrative verification status. Passed: 18 requirements; Unresolved: 2 requirements.
Illustrative verification status. Original Academy diagram using illustrative values; not a measured hardware result.
Read the diagram as a data table
Values used in the illustration
Condition or componentrequirements
Passed18
Unresolved2

The calculation

verified_fraction = passed_requirements / total_requirements

This fraction is a project-tracking measure, not a safety score. Requirements differ in importance and unresolved critical items cannot be averaged away.

Worked example

Illustrative numbers. Replace them with your measured inputs.

If 18 of 20 requirements pass, the verified fraction is 90%. If either remaining item concerns a dangerous power-loss condition, the robot is not ready for release despite the high percentage. Report the unresolved items by name.

Try it step by step

  1. Give each requirement an identifier, measurable acceptance condition and a planned test or analysis method.
  2. Record hardware, firmware, calibration and test-environment versions with the evidence for every result.
  3. Test abnormal conditions and recovery deliberately under an approved procedure, preserving raw observations and unresolved issues.
  4. Release only after required reviews are complete, and define which changes trigger recalibration, retesting or renewed safety assessment.

How to check the result

A reviewer should be able to trace a requirement to its raw evidence and reproduce the configuration that produced the result.

Common mistake to avoid

A checklist cannot certify itself. Independent expertise, applicable standards and legal obligations remain necessary where the real application requires them.

Reference reading

Primary references for the underlying models, APIs or application context. The worked numbers and plots above are educational calculations, not results reported by these sources.

Read our methods, limitations and safety notes.