Target job
Pre-entry reconnaissance, obstacle and foreign-object marks, crack/leak suspects, thermal screening, weak-link maps. Output should tell the crew whether they can enter, from where, where to stop, and which segment holds the evidence.
INSIDE ONLY · NOT FACADE UAV
Enter closed infrastructure from the inside
The owner confirmed that a field application exists. Scene capabilities on this page remain R&D targets: concept model, laboratory rigs and field tasks stay distinct. Target jobs are pre-entry reconnaissance and segment-level evidence in large pipe tunnels, drainage, utility corridors and industrial passages.
Scene selection answers one question: can we repeatedly collect recoverable, aligned, reviewable evidence without sending people into the unknown. Regulatory complexity, medium hazard and recovery certainty set the order — not a marketing “must-have” label.
The first beachhead is therefore de-energised, dry, non-explosive segments with a designed recovery path. Pressurised gas, fully submerged pipes, explosive atmospheres and unrecoverable production lines stay excluded. A robot is not a permit shortcut.
01 · LARGE PIPE TUNNELS
Large pipe tunnels
More space is not more safety. Collapse, standing water, gas layering, weak light and long loss-of-link distances all appear. Flight value is a spatial and atmospheric snapshot before people go down.
Why not an outdoor UAV
GNSS is unstable inside. Contact is often unavoidable. Recovery must be designed first. Open-air obstacle logic fails when “avoid everything” means the wall inspection never happens.
Validation order
G1–G2 on a straight rig and known bends first, then a structurally stable, dry, de-energised tunnel. Water, silt and unknown gas need a separate safety review.
We do not promise
No multi-kilometre continuous coverage, no substitute for structural certification, no work where the machine cannot be recovered.
02 · DRAINAGE
Underground water and drainage
Public network scale shows real maintenance demand. Real demand does not put the first build into a full sewage pipe.
CCTV crawlers remain the mature answer on long, wet, standardised runs. Collision-tolerant flight is more likely at shaft transitions, local obstacles, or where three-dimensional context is missing. PipeGuardian is a scout and gap-filler. It does not declare crawlers obsolete.
Biofilm, wet-wall reflection, H₂S and methane layering, and decontamination chemistry all change sensor validity. Those factors belong in the test plan. Chinese working envelope: drainage use case.
03 · UTILITY CORRIDOR
Utility corridors
Corridors are a natural evidential scene: assets are coded, repeat inspection has value, and recovery paths can be designed.
Data object first
Crews need “which segment, which utility, what is suspect, who reviewed”. PG-Atlas keys on segment and asset codes.
Multi-trade coexistence
Power, comms, water, heat and fire may share a gallery. Thermal, gas and structural findings belong to different trades. The system supplies evidence; it does not overrule specialists.
04 · INDUSTRIAL
Industrial passages
Plants often have stricter HSE. That constraint is an ally: it forces abort conditions, gas thresholds and responsibility into the mission card.
Industrial passages may involve heat, dust, chemical residue or electrical classification. “A UAV can enter” is not a go decision. Medium, gas, Ex zoning and recovery are reviewed case by case. Fail that review and the offer is a bench demo or a non-hazard corridor trial.
PG-X fire or insulation payloads are especially sensitive here. Without independent tests, certification and site approval they stay off the live hazard. Chinese working envelope: industrial use case.
Scene label: A field application exists. That does not publish site names, mileage or an acceptance pack. This page describes target applications. Scene capabilities wait on evidence gates. Named joint trials need written authorisation.