Geospatial Intelligence Platform

The problem

The data is no longer scarce. Fusion is.

Movement in the air, at sea and in orbit, the weather, active hazards, critical infrastructure and what cameras see at street level are each watched by a different system, on a different clock, in a different frame of reference. Connecting an event at sea to the aircraft above it and the camera nearest to it means switching between applications and between analysts. Rows and tickers were designed for a world where data was scarce. The scarce thing now is a single picture.

Separated by domain

Each kind of signal arrives in its own system with its own data model, so relationships between them are found by people, late, if at all.

Separated by clock

Feeds update at very different cadences. Without a common timeline, a picture assembled from them mixes moments that never coexisted.

Separated by frame

Positions arrive in different coordinate systems and conventions. Until they share one reference, nothing can be measured across them.

One surface

A continuous model of the world, at every scale.

FIG. 1Air routes, shipping lanes, satellite orbits at their propagated altitudes, a storm forecast cone and active hazards, composited on one globe. Orbits passing behind the Earth are drawn faint. Tracks are invented for illustration.

S-1

Orbit to street level

Photorealistic terrain and city geometry, explorable continuously from orbital altitude down to a single street corner, with every layer live at every scale.

S-2

Placed on the ground truthfully

Every entity is aligned to the terrain beneath it: aircraft on aprons, cameras on corners, infrastructure at grade, satellites at their true propagated altitude.

S-3

Headings that hold

Each aircraft and vessel is drawn along its real heading from every camera angle, corrected frame by frame, so rotating the view never rotates the traffic.

S-4

One timeline

Every layer shares one clock, with history, scrubbing and playback, so the past state of the whole picture can be replayed together.

Movement

Everything that moves, in one frame of reference.

M-1

Aviation

Global live traffic with full telemetry: altitude, heading, speed, climb and route history. Aircraft are drawn by class, airport ground movement is followed gate to runway, and any contact's recent history is rebuilt as a three-dimensional trajectory.

M-2

Maritime

Global vessel tracking with type, destination, draft and navigational status, drawn with wakes, and handed off to the nearest ground-level view when a vessel comes in sight of one.

M-3

Orbital

The tracked satellite population, classified by mission and propagated with real orbital mechanics rather than approximated positions. Passes over any point on the ground are predicted, and launches are reconstructed along their trajectories.

M-4

Surface transit

Public and commercial vehicles across metropolitan feeds, drawn by mode, with the recent path of any selected vehicle.

Environment & infrastructure

The conditions the movement happens in.

E-1

Active hazards

Fires classified by intensity and seismic events as they are reported, each linked to the nearest ground-level view for visual confirmation.

E-2

Weather

Animated forecast winds, observed radar, cloud cover and lightning on the shared timeline, and tropical cyclone advisories with forecast cones and their uncertainty.

E-3

Critical infrastructure

Data centers, dams, power, submarine cable routes and installations, spatially indexed so they can be filtered, counted and related to anything else on the surface.

E-4

Spatial questions

Which cables land on this coast, which dams lie upstream of this town, what lies within a given distance of a point: answered by the index, not by a search through lists.

Ground level

Cameras as geometry, not as a wall of video.

FIG. 2A street-level plan: cameras at corners, the area each can actually see computed around the buildings that block it, and a vehicle handed from one camera to the next. Dashed where no camera can see. Invented layout.

C-1

Projected into the scene

Public ground-level camera feeds are placed in the three-dimensional scene where they stand, projected onto the geometry they look at rather than embedded as flat video.

C-2

Coverage and blind spots

What each camera can see is computed against the buildings and terrain around it, so the gaps between cameras are as visible as the coverage.

C-3

Calibrated by hand

A camera's orientation and field of view can be adjusted directly in the scene, refining its estimated coverage as it is checked against its own picture.

C-4

Handoff

From any tracked aircraft, vessel or vehicle, the nearest camera with a line of sight is found and brought up.

Ingestion & provenance

Nothing invented, nothing silently stale.

FIG. 3Every feed passes through a pipeline of its own, and every pipeline ends in the same place: one reference in space and time, with the age and source of every position carried alongside it. The scene and the answers to questions are drawn from that one state.

I-1

A pipeline per feed

Each domain is a self-contained pipeline that adapts its source's protocol, whether streamed, polled or delivered in bulk, without disturbing any other.

I-2

Cadence under control

Each feed keeps its own update interval, rate limits and caching, so a slow or rationed source never holds up a fast one.

I-3

One spatial reference

Every position is resolved into a single frame of reference before it is drawn, measured or compared.

I-4

Motion between reports

Smooth movement is reconstructed between discrete position reports by dead reckoning, and corrected the moment a new report arrives.

I-5

Health watched continuously

Every feed's latency and gaps are monitored as it runs, and a failing source is isolated without taking the rest of the picture down with it.

I-6

Provenance on everything

Every entity carries its source, the time of its last update and the health of its feed. When data is late the system says so, and when a feed goes dark its entities age out visibly instead of persisting as stale truth.

Asking the world

Questions answered against live state, with the evidence attached.

Q-1

Aware of the scene

The interface knows what is on screen: position, altitude and bearing of the view, the layers that are active, the entities in sight and the places around them.

Q-2

Interrogate any entity

Select an aircraft, vessel, satellite or installation and ask about it: live telemetry, classification and its relationships to everything nearby.

Q-3

Quantitative answers

How much traffic is above a given level in a sector, which vessels are heading for a port, the largest active fire within a radius: counted against the live layers, with a warning whenever a feed behind the answer is delayed.

Q-4

The whole surface by voice

Layers, camera, sensor modes, tracking and annotation can all be driven by speech, hands-free, as well as by hand.

Q-5

Measure, mark and route

Annotations follow real administrative boundaries rather than drawn approximations. Distances are measured between points or named places, and routes are computed along real road networks and flown through in three dimensions.

Q-6

Seen, not guessed

At street level the interface reads signs, buildings and landmarks from the view itself, and declines to name anything it cannot see.

Sensing & briefing

From the live picture to the finding.

B-1

Sensor modes

The whole picture, terrain, entities and overlays together, rendered as night vision, thermal palettes, enhanced contrast or low light, with live data staying live in every mode.

B-2

Detection overlay

Entities on screen identified with bounding boxes and labels at a chosen density, and a heads-up display of telemetry in a briefing format.

B-3

Layouts for the task

Arrangements tuned for watching, for analysis and for presenting the same picture to a room.

B-4

Scenes as records

The view, the active layers, the sensor mode, the tracked entity and every annotation are saved as one shareable scene, and authored camera tours turn a finding into a briefing that can be replayed.

Connections

The same ideas, elsewhere in our work.

X-1

By invitation.

Every engagement begins with a conversation and is scoped before work begins.

Access

Institutions that need one picture across domains, by invitation.

Subject: Access

Custom builds

Fusion of your own feeds, sensors and assets into the same surface.

Subject: Build

Data partnerships

Providers of feeds, imagery and geographic data.

Subject: Partnership

Investment

Briefings and a demonstration, under confidentiality.

Subject: Funding

contact@carbonyl.org

Geospatial intelligence platform