FEDER Comunitat Valenciana 2021-2027
Proyecto Estrategico en Cooperacion

NeuroAI4Space

Neuromorphic AI Processor for Earth Observation in Space

Pioneering Neuromorphic Intelligence for Space Applications

Developing a low-cost, low-power neuromorphic AI processing board using hardened COTS components for autonomous Earth Observation from smallsats.

TRL 2
Starting Point
TRL 4
Target Goal
Research Objectives

Strategic Goals & Methodology

A comprehensive approach to developing neuromorphic AI processing for space applications through innovative research and technological advancement.

Project Overview

NeuroAI4Space develops a complete neuromorphic AI processing platform for autonomous Earth Observation from smallsats. The project spans from an IP core for Spiking Neural Network (SNN) acceleration to a compact processing board integrating RISC-V and neuromorphic capabilities, complemented by comprehensive hardening methodologies and radiation testing campaigns for LEO orbit qualification.


Neuromorphic Processor

IP core and microchip for Spiking Neural Network (SNN) acceleration with ultra-low power consumption, optimized for on-board satellite inference.

Processing Board

Compact single-board computer integrating RISC-V general-purpose processing with neuromorphic IP for space-qualified Earth Observation payloads.

Radiation Hardening

Hardware and software fault tolerance techniques for COTS components in Low Earth Orbit, including SIHFT and selective redundancy approaches.

Validation & Testing

Radiation testing campaigns, fault injection tools, and verification following ESA standards for space-grade qualification.

Consortium

Project Partners

Four partners combining industrial expertise and academic research to advance neuromorphic AI for space applications.

Coordinator

Intigia S.L.

SME — Petrer, Alicante

Neuromorphic IP Core and microchip development. Specialized in neuromorphic computing architectures for edge AI applications.

SME

EMXYS

SME — Space Electronics

Processing board design, system integration, and verification & validation. Specialist in space-qualified electronics and satellite subsystems.

University

Universidad de Alicante

UA — Software Fault Tolerance

Software-Implemented Hardware Fault Tolerance (SIHFT) techniques and radiation testing campaigns. Researchers: Sergio A. Cuenca Asensi, Antonio Martinez Alvarez.

University

Univ. Politecnica de Valencia

UPV — Hardware Fault Tolerance

Hardware fault tolerance, selective redundancy techniques, and fault injection tools. Research group: STF (Sistemas Tolerantes a Fallos).

Work Plan

Work Packages

Nine coordinated work packages spanning 27 months, from systems engineering through integration, validation, and dissemination.

PT1 Project Coordination
Intigia M1–M27
PT2 Systems Engineering
Intigia M1–M4
PT3 Processing Board
EMXYS M1–M15
PT4 Neuromorphic Processor
Intigia M4–M27
PT5 HW Fault Tolerance
UPV M4–M27
PT6 SW Fault Tolerance
UA M4–M27
PT7 Integration
EMXYS M9–M27
PT8 Verification & Validation
EMXYS M9–M27
PT9 Dissemination
UA M6–M27
Research Output

Publications & Intellectual Property

Planned research dissemination and intellectual property outputs from the NeuroAI4Space project.

Dissemination Plan

Publications coming soon. The project will target high-impact venues in radiation effects, space technology, and neuromorphic computing.

Target Venues

RADECS — Radiation Effects on Components and Systems
ACCEDE Workshop — Application-specific Computing
IAC — International Astronautical Congress
SpaceTech Expo Europe

Intellectual Property

Planned IP outputs covering neuromorphic processing architectures, hardening methodologies, and space-qualified board designs.

Neuromorphic IP Core

SNN acceleration IP for FPGA/ASIC implementation

Hardening Tools & Methods

SW/HW fault tolerance libraries and design methodologies

HIREST Research Group

High Reliability and Safety Technologies