onEDGE · On-board AI computer

An on-board AI computer for uncrewed systems

onEDGE unifies purpose-built hardware, a hardened Linux and hardware-accelerated computer vision in one reference design. Perception, tracking and guidance run on the aircraft itself and keep working when the ground station is out of reach.

The bottleneck

The old choice: a closed platform or years of integration

Closed, vertically integrated platforms are ready out of the box but cannot be audited, modified or extended. The alternative has been to build low-latency video drivers, guidance algorithms and flight-stack integrations from scratch.

onEDGE gives integrators the plug-and-play readiness of a turnkey system on an open, modular and transparent framework. Your team starts from a tested stack and spends its time on the mission. To keep the configuration of the whole drone in one project, see Drone Platform.

Architecture

A separate computer for intelligence, a real-time controller for flight

The companion computer sits above a real-time flight controller and talks to it over a dedicated override bus. The flight controller keeps its real-time loop, and the companion computer can be updated without touching it.

  • Works with ArduPilot, iNav and Betaflight stacks
  • Fixed-wing and multirotor airframes
  • Native integration with Tyko Tech flight electronics

Why onEDGE

What is different about onEDGE

01

Purpose-built hardware

A custom board PC designed for low size, weight and power, with dense industrial IO headers for clean multi-sensor integration and an IP64 enclosure.

02

Intelligence on the airframe

Object detection, tracking, computer vision and autonomous guidance run on board, so the aircraft keeps working when the ground link is slow, jammed or gone.

03

Kernel-level video pipeline

A hardened industrial Linux with our own driver work for low-latency HD digital video and high data throughput under heavy processing load.

04

Open and auditable

Source, kernel and drivers are open to your team to inspect, modify and extend.

05

Scales with the mission

A dedicated expansion lane takes a GPU module such as NVIDIA Jetson for heavier on-board machine learning, without redesigning the airframe.

06

Hardware-rooted trust

Software validation tied to a hardware chain of trust protects the platform against tampering and reverse engineering.

Applications

What operators use it for

Security & defence

Navigation without satellite positioning

Autonomous navigation and on-board perception for missions in contested, GPS-denied or electronically degraded environments.

Public safety

Rooftop-dock response

Aircraft launched from rooftop docks reach an incident before ground teams and stream live thermal and HD maps to the command centre. Mesh-coordinated groups split search areas and close gaps when one unit drops out.

Energy & infrastructure

Inspection of live grids

Visual, thermal and LiDAR inspection of live grids, wind farms and pipelines, and data for digital twins that reveal wear before it becomes failure.

Agriculture

Spot spraying from multispectral data

Multispectral sensing processed on board drives spray valves in real time: treatment where crops are stressed, nothing where they are healthy.

Specification

What is on board

  • ComputeCustom low-SWaP board PC; optional GPU module (NVIDIA Jetson class) on the expansion lane
  • Operating systemHardened industrial Linux with modified kernel and video drivers
  • VideoUltra-low-latency HD digital streaming
  • AutonomyComputer vision, object tracking, auto-piloting algorithms, real-time sensor fusion
  • InterfacesUART / SPI to flight controller; I²C / SPI sensors; industrial IO headers
  • Flight stacksArduPilot, iNav, Betaflight
  • AirframesFixed-wing and multirotor
  • EnclosureIP64
  • SecurityEncrypted software validation with a hardware-rooted chain of trust
  • DeploymentSoftware and data stay under your control, including in your own private cloud

Engineering services

Adapted to your payload and algorithms

Requirements differ from mission to mission. Our engineers adapt onEDGE under a dedicated statement of work.

  • Payload optimisation and integration
  • Secondary sensors: LiDAR, mmWave radar, thermal LWIR cores
  • Client-specific guidance and flight-behaviour algorithms
  • Per-device or tiered licensing of the software stack

Next step

Building an aircraft that has to think on board?

Tell us about your airframe, payload and operating environment. We will show where onEDGE fits and what we would adapt.