A robot is a battery product with sensors, firmware and a cloud account.
We run sampled units through the same fixed course in every factory, so results are comparable between suppliers, lots and re-inspections.

What the checklist misses.
Navigation faults missed by a bench test
Sensor response that changes between units
App pairing or cloud-region failures
Firmware that differs from the approved build
Checks tied to evidence.
| Check | Method | A fail looks like |
|---|---|---|
| Fixed test course | Run a defined route and log completion, interventions and elapsed time. | Missed zones, collision, recovery failure or result outside tolerance. |
| Sensors and controls | Exercise obstacle, edge, lift and applicable safety sensors. | Late response, inconsistent trigger or unsafe state. |
| App and cloud | Pair, register and execute core commands in the destination region. | Pairing failure, wrong region, stale account or unavailable function. |
| Firmware and serial | Capture versions and identities, then reconcile with factory records. | Unapproved build, duplicate serial or incomplete registration trail. |




A result your systems can act on.
Serial-linked evidence, reviewed by a person, delivered as a signed result event and PDF report.
See the sample report- Sample
- 125 units
- Nominal
- 10,000 mAh
- Median
- 9,842 mAh
- Flagged
- 0
Related requirements
The scope references the buyer's approved specification and applicable destination-market machinery, radio and connected-product requirements. Exact requirements depend on product and market.
Not included
Cybersecurity penetration testing, formal safety testing and certification remain specialist partner-lab work.
Clear limits. Clear evidence.
Does inspection guarantee a defect-free lot?
No. Inspection reduces risk. It combines sample evidence with verified factory data and narrow machine checks, but no inspection removes all product risk.
Where can the inspection take place?
At the factory during production or before shipment, or at a neutral warehouse when independence of custody matters more.
Who decides pass or fail?
A qualified person reviews the evidence and makes every result decision. Software compares documents, triages anomalies and drafts the report; AI does not issue verdicts.