CLAIM ONLY WHAT WE CAN REPEAT
This page publishes boundaries, gates and claim labels. The owner confirmed that a field application exists; concept, lab and field stay distinct. Demo operating numbers from a handover pack are excluded. R&D targets need dated tests, conditions, raw records and a reproduction method. The field evidence pack is unpublished.
WHY THIS PAGE EXISTS
If the website, pitch deck, business plan, press and subsidy filings each write their own numbers, the programme loses credit before a reviewer opens the pack. This page is the only place a claim is allowed to upgrade.
Hardware programmes rarely fail because “it cannot be built”. They fail because a target is spoken as a fact too early. Confined-space inspection is especially easy to over-claim: a concept render, a competitor datasheet and a short lab clip can be misread as a validated product. PipeGuardian therefore locks claim rights behind G0–G6. Without the matching gate, pages may only say “target / planned / candidate”.
The same method serves HK–SZ filings. PSTS-class trials want acceptance metrics. ESS / TSSSU want R&D substance and auditable cost. Qianhai matching wants evidence of a Hong Kong award already won. If public copy already wrote mileage and customers that do not exist, every later application is contradicted by the website.
CLAIM LABELS
One fact base serves the site, slides, business plan, press and subsidy filings, so materials cannot fight each other.
VERIFIED
Date, conditions, raw records, reproduction method, named owner. Facts may be stated with their envelope. Example: “ten collision recoveries on this rig, this diameter, this software version”. Not “collision-proof, in production”.
SOURCE FACT
Official statistics, customer documents or public product pages, cited with link and retrieval date. Hong Kong DSD network length, a Flyability manual restriction, an S300 official-page figure belong here. They are not PipeGuardian achievements.
TARGET
Must read “target / planned / candidate”. Cannot be rewritten in the present tense. Diameter ≥600 mm, 18-minute mission endurance and 0.2–3 m wall-hold live here until tests close.
ASSUMPTION
Pricing, volume, conversion, valuation and subsidy forecasts. Sensitivity required. RMB 20M pre-money is a conditioned suggestion, not a traded price.
CONCEPT · LAB · FIELD
The owner confirmed that a field application exists. That confirmation does not turn a concept image into a production photo, a bench log into customer acceptance, or field use into published performance.
CONCEPT
Architecture, concept model, target specifications, exclusion list. Captions keep “not a production photo”. Imagery explains cage and sensor windows only.
LAB
Impact, straight-pipe, bend, calibration and gas-wash records. Maps to G1–G4. In-pipe flight stills without a test ID stay offline.
FIELD
A field application exists. Site names, mileage, availability, customer acceptance packs and a G5 pass remain unpublished. Named pilots need written authorisation.
PRODUCT MATURITY
The concept model supports S0/S1 expression only. A confirmed field application does not auto-upgrade S3 to verified, and does not write G5 as passed. The field evidence pack is unpublished. Until detailed CAD, prototype, tests and customer acceptance exist, visual images are not finished-product evidence.
| Stage | Definition | Current | Upgrade evidence |
|---|---|---|---|
| S0 | Form and mission direction | Set | Scene, constraints, concept decision |
| S1 | System and module architecture | In progress | ICD, mass/power/CG budget |
| S2 | Detailed design freeze | Incomplete | PDR/CDR, CAD, DFMEA, rig |
| S3 | Alpha engineering prototype | Unverified | Collision, straight, bend, loss-of-link, recovery logs |
| S4 | Productisation and trial | Not started | DFM, certification, customer/third-party acceptance |
G0—G6
The 18-month aim is to reach G5. G6 productisation is defined with the actual scene and the certification body. Gates may be prepared in parallel. Public claims may not skip a gate.
| Gate | Question | Minimum pass evidence | What may be said if not passed |
|---|---|---|---|
| G0 | Is the job real and bounded? | 5–8 interviews, 3 site flows, requirements baseline | Problem hypotheses; not “customer-validated” |
| G1 | Can structure hit, be repaired, spare the rotors? | Multi-angle impact, 10-minute module swap target, failure list | Concept cage; not “collision verification complete” |
| G2 | Stable in-pipe, abort on loss of link? | 30 m straight + bends, flight logs, reproduction script | Rig plan; not in-pipe performance |
| G3 | Can data sync, calibrate and trace? | Dictionary, calibration, time and segment index | Information architecture; not “a digital twin already exists” |
| G4 | Is the assist model honest? | Held-out set, recall/false-alarm intervals, misses | Method; not a field accuracy figure |
| G5 | Is the customer process actually better? | Paired tasks, third-party/customer acceptance, value metrics | Joint-trial design; not scaled savings |
| G6 | Can we manufacture and comply? | Reliability, FMEA, maintenance, electrical/radio/battery files | Applicable codes; not “certified / in production” |
DO NOT PUBLISH
Those numbers come from a handover-pack operating demo, not an auditable ledger. Re-introduction needs the original books first.
MINIMUM TEST RECORD
SENSOR CLAIM
Low-light / night vision verifies surface visibility. Thermal verifies temperature difference and heat sources. Gas sensing verifies the target gas. Fusion can add context. Channels cannot substitute for each other. See the insight article Methane sensing versus thermal imaging.
PG-X GATE
Fire-suppressant and insulation modules do not enter a real hazardous scene until media, throw, false trigger, contamination, explosion protection, fire service, insulation and duty-bound evidence are complete. An interface is not a capability.
HOW WE CITE OTHERS
Competitors, papers, government public data and codes may be cited. They must be traceable, and they must not be rewritten as PipeGuardian results.
Allowed: source, retrieval date, envelope — for example “the Flyability manual states the original edition is not for explosive / flammable atmospheres”, “Hong Kong DSD discloses underground-drain scale”. Not allowed: copying Elios, Scout 137, S300 or THS-04 endurance or accuracy onto the PipeGuardian spec sheet, or writing a review-paper lab metric as product capability.
Roadmap, specifications and safety inherit labels from this page. An unmarked present-tense performance sentence on any page is a defect, and it is fixed before new features.
This page also binds images. A concept model may only explain protection and window semantics; the caption keeps “not a production photograph”. Rig photos must point at a test ID. Any “in-pipe flight” still stays offline until G2 logs are archived. If press needs a picture, prefer an authorised portrait and a labelled concept image over a stock drone.