ArchiMetis

Discovery

Project Snapshot

Status

On-going

Start Date

2026

End Date

2027

Programme

Discovery

Partners

OHB Hellas role

Prime

The Objective

ARCHIMETIS is an ESA-funded activity led by OHB Hellas, focusing on the evaluation of neuromorphic computing for on-board space applications. The project investigates the use of spiking neural networks (SNNs) deployed on dedicated neuromorphic hardware, with the aim of enabling low-power, real-time processing under space constraints.

The activity centres on the SENNA neuromorphic processor, a European architecture designed for event-driven computation and ultra-low-power inference. ARCHIMETIS aims to assess its suitability across representative mission use cases, covering different categories of onboard processing such as control-support functions, data handling and compression, and Earth observation data prioritisation.

A structured and staged benchmarking approach is followed, starting from comparable baseline models and progressively increasing model complexity. This allows a consistent evaluation of latency, energy consumption and accuracy across multiple platforms, including both neuromorphic and conventional embedded AI solutions. The results will support the identification of suitable processing strategies for future space missions.

OHB Hellas Contributions

OHB Hellas leads the ARCHIMETIS project and is responsible for the overall coordination and system-level definition of the activity.

The main contributions include:

Definition and selection of mission-relevant use cases based on previous OHB Hellas activities and ESA priorities
Specification and development of baseline AI models tailored to embedded and neuromorphic execution
Adaptation of models to hardware constraints, including quantisation and architecture simplification
Cross-platform benchmarking across SENNA, BrainChip Akida, Coral Edge TPU and NVIDIA Jetson
Analysis of performance trade-offs in terms of latency, energy consumption and accuracy

In addition, OHB Hellas contributes to the consolidation of a consistent evaluation framework for on-board AI processing, supporting the maturation of European capabilities in neuromorphic computing and embedded intelligence for space applications.