Counter-UAS

The Future of
Spatial Awareness

A decentralized sensor mesh for early detection and real-time tracking of airborne objects.

01The system

A tracking mesh, woven from ordinary cameras.

LocalEyes is a decentralized 3-D tracking mesh for the low-altitude sky. Our localization algorithm fuses overlapping observations from many low-cost commodity cameras, each one a simple local observer, into a single real-time picture of where airborne objects are, where they are heading, and how they are moving. No single sensor sees everything. Together, they see it all.

It began with a tree. From one viewpoint, two branches appeared to intersect but there was no way to know whether they truly did without a second local eye. That became LocalEyes: many simple observers, cooperating to reconstruct the true structure and motion of a scene.

Local eyes, localizing local skies.

02How it works

Deterministic by design.

Each node detects motion in its own view. The mesh fuses those observations across nodes and resolves them into 3-D position, direction, and trajectory in real time. Because the core is purely geometric, it runs on inexpensive commodity hardware. No specialized radar, no centralized sensing stack.

Sensor-agnostic

Works with ordinary RGB cameras and extends to other sensing modalities chosen based on application.

Geometric core

Localization is deterministic geometry, verifiable, repeatable.

Real-time

A C++ / CUDA and Rust core engineered for low latency, scaling smoothly to large sensor meshes.

Decentralized

Every node contributes local information to a shared estimate. Coverage grows by adding nodes, not by replacing infrastructure.

03Validation

Measured against Vicon motion capture and a Leica MS60 total station.

Geometry baseline

Single drone

~5.3 cm path-following error over 39 m, indoors against Vicon (4 cameras).

Three at once

Multi-drone

Three drones at once, kept separate ~6.1 cm across flights of ~76-110 m, indoors against Vicon (4 cameras).

Early field result

Outdoor

First open-field run: ~9.1 cm over 6.4 m against a Leica MS60 (2 cameras). A short-range proof longer-range validation is in progress.

04Deployment

Three ways to cover a site.

01

Upgrade packages

LocalEyes-compatible camera modules for GPS-networked sensor systems already deployed across large outdoor sites turning existing monitoring infrastructure into airspace awareness.

02

Standalone nodes

Rugged, self-calibrating sensor units with onboard compute, networking, and GPS. Placed around a site in hours to create ad-hoc aerial coverage.

03

Software licensing

Pure software for existing CCTV and security-camera networks. Where overlapping coverage already exists, LocalEyes produces 3-D tracks from the cameras you already own.

Applications
Industrial sitesPorts & terminalsPower plantsAirportsBorders & perimetersLarge events
See the use cases
05Team

Built in Toronto by three engineers.

Souren

Robotics PhD candidate at the University of Toronto working on multimodal perception and embodied systems. Previously built LiDAR-calibration tooling in C++/Qt and Python/OpenCV and shipped hardware-software products end to end. Leads research, architecture, and integration.

Mahyar

Systems engineer in production C++ and Rust. Maintained national-scale weather and air-quality data pipelines at Environment and Climate Change Canada, and built a satellite-network routing simulator from scratch in Rust. Owns the real-time core and networking.

Gibran

Mechanical and automation engineer (M.Eng). Designed and validated industrial automation systems as a consulting engineer at Lincoln Electric; autonomous-flight work with UofT drone racing. Owns actuated mounts, calibration mechanisms, and deployable hardware.

06Contact

Tell us about your airspace.