4D-360-AI™

Video gallery

The full 4D-360™ pipeline, end to end.

Short clips covering AI driven Video/Lidar field data capture and feature extraction, advanced GIS integration and Conflation (Utility Network Adjustments).

4D-360-AI Conflation: 100% Automated 2:19

4D-360-AI Conflation: 100% Automated

An existing land base, drawn decades ago from many sources, sits several feet off highly accurate lidar. Five automatic steps put it where it belongs: extract walls and road edges straight out of the lidar; match features in both, with every matcher also tested on deliberately shifted data; solve exact shift vectors; conflate the land base onto the survey, with a rigorous least squares adjustment that keeps its angles, distances and utility network shapes; and check the result against quality gates. About 27 seconds per square mile, with no clicks.

Conflation for GE Vernova Smallworld Geo Network Management: Put the Data Where It Belongs 3:08

Conflation for GE Vernova Smallworld Geo Network Management: Put the Data Where It Belongs

A gas network in GE Vernova Smallworld Geo Network Management, digitised many years ago, sits several metres from its true position. Five steps put it where it belongs, without leaving the tools the custodian already uses: draw the survey area and extract it; build the shift model from known vectors, with the engine’s fit statistics back in seconds; review the run in the browser — green for where each feature was, purple for where it goes — and approve it against a name; apply on the desktop into the master and every design that touches the area, each under its own checkpoint, reversible until it is published; and read on the status tab what was written, when, and from which manifest.

Smallworld-AI — Electric Office: Ask Your Network in Plain Language 12:34

Smallworld-AI — Electric Office: Ask Your Network in Plain Language

A working session inside Smallworld Electric Office, with Smallworld-AI in the application itself. Plain-language questions answered from the live database: a downstream trace painted onto the map with the platform's own trace display, a true-scale page plotted off your own circuit templates, and the application's own dialogs and Quality Manager read back in plain language. Nothing pre-scripted — and anything that would leave the machine asks first.

Smallworld-AI — Electric Office: One Outage, Two Deliverables 5:01

Smallworld-AI — Electric Office: One Outage, Two Deliverables

A planned outage worked end to end in Smallworld Electric Office. A low-voltage breaker has to come out for maintenance: one question traces everything that goes dark, across all three phases, paints it on the map in the platform's own symbology, hands customer services the service points to write to, and plots the contractor a scale drawing off a template already in the install. Normally about ten steps across two tools.

Smallworld-AI — Electric Office: Build It, Then Complete the Record 4:04

Smallworld-AI — Electric Office: Build It, Then Complete the Record

An overhead line extended to reach new load at the end of a feeder, and the record of it completed the same day, inside Smallworld Electric Office. Two words spoken out loud open the layout wizard the planners already have — nothing is built by saying it — and the span and its poles are placed, then completed. The wizard leaves the record without voltage or usage; a plain-language instruction sets three fields at once, checked against the collection’s own data model before anything is written. The assistant proposes; a person presses Apply, then Update; the write lands in their own design, under their own permissions.

Smallworld-AI — Physical Network Inventory: What We Own, and Where the Fault Is 11:39

Smallworld-AI — Physical Network Inventory: What We Own, and Where the Fault Is

Telecom network questions asked in plain language against a live Smallworld Physical Network Inventory. How much plant do we own, and how it was built. A fault along a fibre turned into a place on the map and a dispatch sheet centred on it. The duct route a crew would actually dig between two cabinets, not the cable path. Then a greenfield estate sized into a bill of materials, on your own design standard.

Smallworld-AI — Physical Network Inventory: Build It, Then Correct the Record 3:52

Smallworld-AI — Physical Network Inventory: Build It, Then Correct the Record

The civils for a new fibre customer’s lead-in, built and as-built inside Smallworld Physical Network Inventory. Four commands spoken out loud — create mode, the catalogue item, start, complete — place a duct route with a chamber at each end, using the application’s own Favourites. Then the record of that same route is corrected where the crew deviated: the template saved it as trench, they bored it rather than open the road, and three fields change together — route type, diameter and centre point depth — checked against the collection’s own data model before anything is written. The assistant proposes; a person presses Apply, then Update; the write lands in their own design, under their own permissions.

Capture 0:29

Capture

Drive a corridor with the roof-rack 360° camera + GPS rig. The footage is the raw material everything else depends on.

Feature Extraction 1:02

Feature Extraction

Turn the drive into point clouds, photo-real scenes, surface meshes, and asset inventories (poles, wires, vegetation).

One Drive, Every Asset Found (Capitola, CA) 0:29

One Drive, Every Asset Found (Capitola, CA)

A drive through Capitola, California, seen in the 4D-360™ explorer. Every tree is outlined to the shape of its own canopy — not a box drawn around it — alongside road signs, light poles, crosswalks, manholes and lane markings, each carrying the confidence it was found with.

Phone Video → 3D Point Cloud 0:49

Phone Video → 3D Point Cloud

A single phone-camera pass around a parked vehicle, fed into the 4D-360™ pipeline. The AI inspects every frame to pick the optimal reconstruction method, then produces a measurable, navigable 3D scene.

Ladder Company 3 — Rebuilt in 3D from a Single Phone Video 0:54

Ladder Company 3 — Rebuilt in 3D from a Single Phone Video

The remains of Ladder Company 3, on display at the September 11 Memorial Museum in New York, rebuilt as a three-million-point 3D scene from one 64-second handheld phone video — no laser scanner, no survey rig, no markers. Orbit it, inspect it from any angle, measure across it.