Radareye

All-weather perception,
without the labels.

A sensor-agnostic radar foundation model. Built from the physics up, learns without labels, and gets better with every datum.

Start a pilot → Read the technical brief
All-weatherNear-zero disengagements where vision-only fails and LiDAR struggles.
Sensor-agnosticRuns on any radar: low-cost, 4D imaging, or bespoke variants. No hardware lock-in.
Self-supervisedEvery collected datum becomes training data. The model improves at near-zero marginal cost.
Software-onlyDeploys as IP license on existing silicon. No new sensors, no manufacturing risk.
Supported by
Funded by UK Government UKRI Innovate UK Advanced Propulsion Centre UK Zenzic CAM Pathfinder NVIDIA Inception Program Pacific Northwest National Laboratory COVE EnSilica
The problem

The one sensor that keeps working in bad weather is the one sensor without an AI supply chain.

Vision fails. LiDAR degrades. Radar keeps detecting, but no one can annotate it at scale. Every new radar generation restarts the labelling bill. We fix this.

The solution

Label efficiency you can price on any radar.

We drastically cut the number of labels needed for your radar perception task, whether that is labelling budget or months of programme time.

We also let you use corpora no "Mechanical Turker" can annotate, such as underwater sonar.

The tech

Self-supervised data flywheel.

Our foundation model is powered by self-supervised learning. We support cross-modal and radar-only pretraining options.

The cross-modal variant is our most capable model and includes advanced features such as visual priors. The radar-only variant removes the paired radar-camera assumption, opening the door to broader applications such as maritime, counter-UAS, and Arctic surveillance.

See how our technology works

Selected engagements
  • 01

    Pacific Northwest National Laboratory (PNNL)

    Scoping under LOI.

  • 02

    UK Government

    CAM Pathfinder and DRIVE35 programmes.

  • 03

    Innovate UK and COVE innovation hub

    Global Incubator Programme: Seabed to Space Dual-Use Technologies, Halifax Canada.