AI-lert Box
AI and alert, in one weatherproof box. Radar for movement, a camera for recognition, an air-quality stack for the environment and a beacon mesh for everything that carries a tag. It does not report the accident. It stops it.
| Component | What it contributes |
|---|---|
| Raspberry Pi | The compute. Inference and fusion run on the box, not in a data centre. |
| Radar | Range and closing speed, in the dark and through weather. |
| Camera | AI object recognition — person, vehicle, tug, cone — with a confidence score. |
| Air quality | Particulates, CO₂ and the gases that say the environment has changed. |
| Beacon mesh | Temperature and humidity across the whole area, plus anything wearing a tag. |
| Dashboard | Time-based graphs, recognised objects and alert history in one view. |
Three signals, one verdict
Radar gives distance
Something is fifteen metres out and closing at eight kilometres an hour. No camera needed for that, and no light either.
Vision gives identity
The camera says what it is and how sure it is. A person behind a vehicle is a different problem from a cone behind a vehicle.
The mesh gives context
Tagged crew, tagged plant, temperature and air quality across the site. Distance plus identity plus context is enough to warn the right person in time.
Aprons, verges and everywhere else people and machines share ground
Aprons, de-icing pads and baggage roads, where ground crew and vehicles share the same metres of tarmac in poor light.
A moving work zone with live traffic beside it. The box travels with the works and warns before the incursion.
The gentler end of the same technology: air quality in the room, occupancy on the floor, and offers that reach the person already at the door.
Apply for the pilot
A limited number of sites in 2026. Tell us the environment and we will come back with a box configuration and a timeline.
We answer pilot requests within two working days.