Validation and Feedback Records

Use this guide after a Physical AI layout, simulation, robotics, or runtime study produces output that may be reused. Validation records keep the scenario trustworthy and auditable.

Feedback Flow

Prerequisites

Requirement	Why it matters
Reviewer	Engineering output needs an accountable decision maker.
Evidence package	Runs, KPIs, exports, and assumptions need stable references.
Feedback path	Field observations and runtime findings must return to the scenario record.

Inputs

Input	Review focus
Run evidence	Scenario ID, run ID, engine, duration, seed, parameters, and output files.
Assumptions	Geometry simplifications, physics parameters, process limits, and missing data.
Decision record	Accepted, rejected, needs more evidence, or reuse-limited status.
Field feedback	Observation, measured difference, issue, owner, and follow-up action.
Revision record	Updated asset version, constraints, semantic labels, or runtime setting.

Procedure

1. Gather the scenario package and run evidence.
2. Review assumptions and known limitations.
3. Compare outputs against engineering expectations or field observations.
4. Record reviewer decision and allowed reuse.
5. Capture feedback from runtime tests, operator review, or field validation.
6. Update asset packages, constraints, or scenario versions as needed.

Expected Output

The output is a validation record that explains what was tested, what evidence was reviewed, what decision was made, and what changed next.

Validation Checklist

• Scenario ID, asset version, engine, and run ID are visible.
• Assumptions and limitations are documented.
• Reviewer decision and allowed reuse are explicit.
• Field feedback is linked to follow-up action.
• Revisions create new versioned records.

Failure Handling

Symptom	Response
Evidence is incomplete	Keep the scenario in review status and list missing records.
Field result diverges from simulation	Record difference and update assumptions or source data.
Reuse scope is unclear	Limit reuse until reviewer defines permitted scenarios.
Revision overwrites prior evidence	Create a new version and preserve the previous record.

Related Docs

• Physical AI
• Simulation and What-If Runs
• Robotics and Runtime Handoff