Quadrion AI/Drone inspection & survey

Four dimensions.One view.

Drone video, 3D models, point clouds, maps and issues, fused into one georeferenced scene with one timeline. Fully offline, on your own machine.

Request a demo Download the free Player

00/The Quarter

Four dimensions. Six real assets.

  • XDistance

    Urban ring road

    7.55 km · 2,115 defects

    Orthomosaic + pavement condition

  • YArea

    Tank terminal

    2.1 × 1 km · 757,579 points

    Photogrammetry cloud + engineering model

  • ZHeight

    Flare stack, 80 m

    78 mapped corrosion patches

    Textured model + corrosion mapping

  • TTime

    Stockpile yard

    2 surveys · 10 days apart

    Two survey dates + volumes

x distance · y area · z height · t time  =  one georeferenced scene

01/Road·X, distance

Everykilometre.

7.55kmurban ring road
2,115mapped defects
87.6network PCI · medium
Patched section of asphalt at the kerb, outlined
High · PatchingPatch, 22.9 m²D0941 · km 3.714 · screening
Alligator cracking across a traffic lane, outlined
High · Alligator crackingAlligator cracking, 64.4 m²D0991 · km 3.804 · screening
Block cracking along a carriageway, outlined
High · Block crackingBlock cracking, 955.8 m²D0920 · km 3.963 · screening

02/Tank terminal·Y, area

The wholesite.

757,579captured points
2.1km × 1kmsite extent
8storage tanks + jetty

03/Storage tank·inside, LiDAR

Insidethe tank.

280,000LiDAR points · 2 cm
6graded findings, with photos
Patch damage on the tank roof plate
Severity 4 · Patch damageTop plate patch damageF02 · Roof · 8.48 m
Coating blister on a vertical shell joint
Severity 3 · BlisterVertical joint, 3 o'clockF09 · Shell · 4.74 m
Crack in the bottom plate of the storage tank
Severity 5 · CrackCrack in bottom plateF01 · Bottom plate · 0.04 m

04/Flare stack, 80 m·Z, height

Up thestack.

80mscanned top to bottom
78mapped corrosion patches
77 · 1moderate · light
Moderate rust on the flare neck bolted connection
Moderate visible rustFlare neck / bolted connectionF04 · Flare head · 76.7 m
Moderate rust along a stack cladding seam
Moderate visible rustStack cladding / seamF44 · Stack / access · 55.3 m

Graded issues by class

    05/Tower facade, 74 m·Z, height

    Everyelevation.

    656graded issues
    715mapped areas
    474 · 182minor · moderate
    Roof canopy fascia with grey run-off streaks, class overlay on top
    Minor · Staining or run-offRun-off on the roof canopy fasciaD159 · North elevation · 64.7 m
    West wall cladding in visible lightWest wall cladding in thermal, a warm spot lower right
    Moderate · Thermal anomalyWarm spot behind white claddingD397 · West elevation · 36.3 m

    06/Stockpile yard·T, time

    Acrosstime.

    76,199m³Survey 1
    63,402m³Survey 2, 10 days later
    −12,797m³Net change · 19 piles
    P0214,687 m³11,379 m³
    P057,496 m³4,825 m³
    P181,314 m³215 m³

    07/In one place

    Video, models, point clouds, maps and issues. Finally in one place. Offline.

    No cloud uploadNo accountYour disks

    08/The product

    One view.

    All six real datasets, live in the app. Pick one on the left, inspect the scene, and every finding sits on the right with its evidence.

    Quadrion AI — GCC inspection programme — Scene
    SceneIssuesVolumesReports Offline AI agent · own key
    Flare stack, 80 mLive WebGL · real data
    AI agent your key · off by default

    Where is the corrosion on this stack?

    Survey 1 Survey 2 +10 d Flights
    1. 01Six real datasets
    2. 02Live 3D scene
    3. 03Findings with evidence
    4. 04One timeline
    5. 05AI agent

    09/Chapters

    From flight
    to handover.

    Eight chapters, 09.1 to 09.8, of the same job, on the same real data. Nothing leaves the machine between them.

    Corrosion, mapped

    Each drone photo is segmented into corrosion classes, then the mask is projected back onto the textured model as a patch you can measure.

    • Patches78 · 77 moderate, 1 light
    • ZonesHead · Stack · Base
    • EvidencePhoto + class mask
    F44 · Stack cladding / seam · 55.3 mMove to compare
    Drone photo of a stack cladding seam with a band of rust
    LightModerateHeavyUncertain

    Every finding, located

    Six graded findings inside one storage tank, each pinned in 3D on the LiDAR scan and backed by the photo it came from.

    • Findings6 with photos
    • Severity1 to 5
    • LocatedIn 3D, to the cm
    Storage tank interior · findings06

    Pavement, screened

    2,115 defects mapped from the 2D ortho along 7.55 km, each graded for screening and rolled up into a PCI every 250 m.

    • Sample units1,783
    • Network PCI87.6 · medium
    • ScaleASTM D6433
    Urban ring road · D0991 · km 3.804PCI per 250 m
    Alligator cracking across a traffic lane, with its mapped outline High · Alligator cracking64.4 m² · screening grade
    0 kmHover a section7.55 km

    Facade, in two spectra

    1,146 RGB and 36 thermal photos of one tower, each issue tied to the area it maps on the model. Heat that hides behind cladding shows up in thermal.

    • Photos1,146 RGB · 36 thermal
    • Graded issues656
    • Uncertain45, for review
    Tower facade · west elevationRGB + thermal
    West wall cladding in visible light, with run-off and rust on the rods
    RGB · 31.8 m · staining, rust on rods
    West wall cladding in thermal, a diffuse warm spot lower right
    Thermal · 36.3 m · warm spot, moderate

    Volumes, surface to surface

    Two surveys ten days apart, 19 piles, cut and fill per pile. Hand the surfaces to Civil 3D, TBC or 12d when the engineers need them.

    • Piles19
    • MethodTriangulated toe
    • ExportsCivil 3D · TBC · 12d
    Stockpile yard · Survey 1 → Survey 2−12,797 m³
    76,199 m³Survey 1
    63,402 m³Survey 2, 10 days later
    Survey 1Survey 2CutFill

    GPU processing

    Optional NVIDIA GPU processing takes photo matching from minutes to seconds. The CPU path is always there.

    • HardwareNVIDIA, optional
    • StagePhoto matching
    • RunsOn your machine
    20×

    faster photo matching

    CPU255 s
    NVIDIA GPU12.9 s

    255 s → 12.9 s  ·  same project, photo matching stage

    An AI agent in every window

    Ask about the project in plain language. It answers from your own data, with your own key and provider, and it is off until you turn it on.

    • DefaultOff
    • KeyYours
    • ScopeThis project
    AI agentOwn key
    Which tank findings need action first?
    F01 photoF02 photo
    Where is the road in the worst condition?
    Provider: your choiceCloud AI: off by default

    Hand over the whole programme

    Export one read-only .aio package. Your client opens it in the free Player, offline, with cloud AI forbidden unless you allow it.

    • Package.aio
    • ViewerFree Player
    • Cloud AIOff by default
    Export package.aio
    .aio
    handover.aioRead-only client package · 6 datasets
    • ├Urban ring road · ortho + PCI7.55 km · 2,115 defects
    • ├Tank terminal · photogrammetry cloud757,579 pts
    • ├Storage tank interior · LiDAR280,000 pts · 6 findings
    • ├Flare stack, 80 m · textured mesh78 patches
    • ├Tower facade, 74 m · model + photos656 issues · 715 areas
    • └Stockpile yard · Survey 1 + Survey 219 piles
    Read-onlyCloud AI forbiddenOpens offline in the free Player

    10/Trust

    Built for restricted networks.
    Locked on, by design.

    These are not settings you have to find. They are how Quadrion AI is built, for teams whose data never leaves the site.

    AI agentAn AI agent sits in every window. It is off by default, and when you turn it on it uses your own key with the provider you choose.

    Bring the whole programme into one view.

    See Quadrion AI on your own data, on your own network. Or send your client a read-only package and the free Player.

    Request a demo Download the free Player

    Windows desktop · Offline-first · Exports to Civil 3D, TBC and 12d

    Direction D3 · The Quarter