Robotics in food production
A cooking cell is a motion problem, a food safety problem and a thermal problem at once. It sits inside our four fields, and we have a sister brand in the group that sells the result.
Where we work
Four fields that share the same physics and the same regulator
They are not four businesses. A cargo cycle, a battery pack, a model that decides when to charge and the board it runs on are one problem seen from four sides, and the same record covers all of it.
01. Electric vehicles
The pack, the charge and the heat
Most electric vehicle projects fail on thermal behaviour or on a charging interface that a standard moved under. Both are visible in the record years before they bite, because someone already hit them and filed about it.
We work on pack architecture, cell selection against a measured duty cycle, the battery management strategy, and the charging interface with the standards that govern it. Type approval and CE are planned from the first sketch rather than discovered in month nine.
The part that surprises people is how early it is decided. Cell format and cooling strategy together fix most of the pack, and both are usually chosen in a week that nobody thought was important at the time.
02. Electric transport
Everything that moves a load and is not a passenger car
Light commercial vehicles, cargo cycles, rail, port and yard equipment. The economics here are decided by the depot rather than by the vehicle, and the grid connection is usually the constraint nobody costed.
We measure the duty cycle instead of assuming it, size the charging behind it, and produce a total cost that survives being read by an accountant rather than by an enthusiast.
03. Artificial intelligence
Models that have to be right, not models that have to demo well
A model in a vehicle or on an industrial device answers for its output. That changes the engineering: the evaluation set has to be yours, the failures have to be enumerated before launch, and the fallback when the model is unsure has to exist in the product rather than in the roadmap.
We build on device and edge inference, we evaluate against data you hold and can rerun, and we settle the EU AI Act role of the system before it ships instead of after a customer asks.
A demo answers to a room. A model in a vehicle answers to whoever is in front of it. Building for the second one means writing the failures down before launch rather than discovering them afterwards.
04. Hardware
A thing you can hold, measure and take to a test house
Schematic, layout, a board that boots, an enclosure that closes and firmware that survives a power cut. The unglamorous half of innovation, and the half that decides whether any of the rest ever ships.
We pre test for EMC and safety before a booked slot at a test house, because a failed slot costs more than the pre test and moves the launch by a quarter.
Adjacent work we take, and why it is adjacent
Energy and storage
Stationary storage is a vehicle pack that stopped moving. The cells, the management and the safety case are the same argument with a different duty cycle.
Controlled environment growing
Sensors, power electronics and a model that decides. Different industry, identical stack, and the record is much emptier than in automotive.
The engineering underneath
Twelve things that appear in almost every project
Cells, heat, a connector, a grid connection, a depot, a board, an enclosure, a chamber, a sensor, a bench, a network and the device it runs on. Different industries, the same dozen problems.
Is your field one of these?
If it is next door to one of them, ask anyway. If it is genuinely somewhere else we will say so quickly.