SETANTA / AI & ONBOARD INTELLIGENCE

Intelligence.
On board.

Turn sensor data into useful information, closer to where it is captured.

From image analysis to event detection, we explore AI workloads around the realities of onboard computing: available power, processing capacity and limited communications.

Discuss an AI workload Explore applications
ONBOARD AIINPUT → INSIGHT
From sensor data to an onboard decision A conceptual sequence: camera and sensor input feeds model inference, which produces events, features or priorities for the application. 01 / OBSERVE Sensor data 02 / INTERPRET Model inference 03 / INFORM THE APPLICATION EVENTSFEATURESPRIORITY
Conceptual workflow · evaluated for each model and target
MODEL + MISSION + HARDWAREENGINEERED TOGETHER

01 / APPLICATION AREAS

Make the data
work harder.

Start with the information your mission needs. Then shape the workload around the system that has to run it.

01

Earth observation
triage.

Identify clouds, coastlines, fire, flooding or changes in imagery to help prioritise frames for review and downlink.

IMAGERY → USEFUL FRAMES
02

Space-weather &
sensor alerts.

Monitor sensor streams for anomalies, threshold crossings and transient events, turning changing signals into concise alerts.

SENSOR STREAMS → EVENTS
03

Optical navigation
& rendezvous.

Extract and match visual features, and estimate pose from camera data, to support navigation processing.

CAMERA DATA → FEATURES & POSE
04

Communications &
downlink priority.

Assess signal quality, compress payload outputs and prioritise data for transmission within constrained contact windows.

PAYLOAD DATA → PRIORITISED OUTPUT

Application areas for evaluation. Model performance, target support and mission readiness are assessed for each configuration.

02 / FROM MODEL TO HARDWARE

A model.
A running FPGA.

Explore the Vertis workflow
DEMONSTRATED WORKFLOW

Vertis has demonstrated its supported automated workflow with multiple computer-vision models on AMD FPGA hardware.

  1. 01 Import model
  2. 02 Compile & verify
  3. 03 Program FPGA
  4. 04 Run inference

Vertis is under active development. Supported models and targets are evaluated individually; hardware execution and flight qualification are separate milestones.

Apply to the closed alpha

03 / ADD HARDWARE TO THE STACK

Have your
own model?

Add hardware to your stack.

Explore our onboard computers and modular Space Tiles. Bring your AI model and system requirements, and we’ll help you choose a hardware configuration to evaluate.

Explore our hardware
DANU modular onboard computerDANU / MODULAR ONBOARD COMPUTE

YOUR NEXT WORKLOAD

What should your
system see?

Tell us about your application, the data it uses and where it needs to run. We’ll discuss a practical next step with your team.

Discuss an AI workload contact@setantaspace.com