Safety

You are here:

Safety

Next‑generation safety systems must be able to anticipate risks, interpret complex situations and operate reliably across a wide range of conditions, from low visibility to emergency scenarios. To address these needs, NeAIxt delivers modular Edge‑AI technologies that run directly on devices and sensors, enabling fast, energy‑efficient and robust solutions designed to protect people, infrastructures and environments.

Through 5 Ecosystems, NeAIxt uses edge AI to detect risks in real time, even in challenging conditions, enabling faster responses, better decision‑making and more resilient safety systems for cities, industries and emergency services.

Ecosystem : Crowd Management

This Ecosystem enables real‑time, Edge‑AI‑driven crowd management by combining on‑device intelligence close to cameras and sensors with 5G connectivity and far‑edge orchestration, to detect anormal movements or risky situations and support safer, smoother decision‑making in large‑scale venues such as stadiums.

	Ecosystem Partner list	Activity details
	Instituto de telecomunicações	Ecosystem and Use case leader: IT-PT develops and deploys the Edge‑AI and frameworks enable real‑time crowd detection, behaviour prediction, routing assistance and overall system integration in stadium‑scale environments.
	Fraunhofer Portugal	FhP AICOS provides the far‑edge orchestration platform, enabling large‑scale IoT device management, dynamic computation distribution, and AI deployment across dense networks.
	MEO	MEO supplies the 5G SA slice for low‑latency connectivity, ensuring reliable communication between sensors, nodes and Edge‑AI applications.
	STMicroelectronics CROLLES	ST-CRO contributes advanced FD‑SOI technology, memory integration and 3D bonding to support high‑performance used for Edge‑AI inference in the crowd monitoring system.
	STMicroelectronics GRENOBLE	ST‑GRE develops the new AI‑enabled microcontroler used in the crowd monitoring architecture, providing on‑device neural processing and performance‑optimized embedded compute.

Ecosystem : Identification in difficult situations

This Ecosystem makes person identification reliable in difficult conditions such as darkness, smoke or fog by combining polarimetric sensing with compute‑efficient Edge‑AI mapped on platform, enabling robust detection and identification by day or night and improving safety and situational awareness.

	Ecosystem Partner list	Activity details
	IRnova	Ecosystem leader: IRnova provides and advances the thermal & polarimetric HD sensors and contributes to system/app integration and datasets.
	Swedish Defence Research Agency	Use case leader: FOI develops the AI models and inspection tools for person detection and identification, targeting compute‑efficient deployment and validated protocols.
	Royal Institute of Technology	KTH ports and maps partners’ algorithms to the platform to achieve large energy‑delay gains and hardware‑ready designs.
	Strikersoft	Strikersoft builds Edge pipelines and components, supporting integration across sensors and applications in the ecosystem.

Ecosystem : Image processing

This Ecosystem brings fast, reliable and high‑accuracy image processing to resource‑constrained embedded devices by adapting and securing optimized AI algorithms on Edge‑AI hardware, enabling applications such as biometric recognition, fraud detection and identity‑document security verification.

	Ecosystem Partner list	Activity details
	Thales	Ecosystem and Use case leader: Thales develops optimized embedded AI models for identity‑document processing, ensuring fast inference, low power consumption, secure model deployment, and compliance with border‑control constraints.
	STMicroelectronics ROUSET	ST RST develops the microcontroler, AI accelerators, and secure embedded features, providing the Edge‑AI hardware foundation for the image‑processing demonstrator.

Ecosystem : Disaster Detection

This Ecosystem enables early and real‑time detection of natural and man‑made disasters by combining high‑performance, energy‑efficient Edge‑AI with satellite on‑board processing, delivering fast analysis of Earth‑observation data directly in orbit to support quicker response in remote and hard‑to‑access areas

	Ecosystem Partner list	Activity details
	Integrated Systems Development (ISD)	Ecosystem and Use case leader: ISD develops the next‑generation computational node and leads the on‑board disaster‑detection pipeline, ensuring real‑time throughput and low‑power operation.
	University of Athens (UOA)	UOA creates firmware for optical and streams, enabling high‑throughput, low‑latency Earth‑observation analytics on the ISD node.
	University of Piraeus (UPRC)	UPRC designs high‑performance, radiation‑tolerant accelerators, optimizing models for real‑time on‑board detection.
	Geosystems Hellas (GSH)	GSH develops neural networks and generates datasets for disaster detection, deploying and validating them on emulated and architectures.
	Klepsydra	Klepsydra provides a hardware‑agnostic, high‑throughput Edge‑AI inference engine and tools to maximize performance and reduce power consumption on the computational node.

Ecosystem : Disaster Management

This Ecosystem combines stationary monitoring systems and AI‑enabled drones running microcontroler‑based inference to detect and localize people in disaster zones, fusing ground and aerial data in real time to improve situational awareness and support faster, safer and more effective emergency response.

	Ecosystem Partner list	Activity details
	TUBITAK	Ecosystem and Use case leader: TÜBİTAK develops the full disaster‑response system by integrating stationary cameras and AI‑equipped drones, implementing human detection on mincrocontroler, secure wireless data transfer, dataset creation, and real‑time multi‑source fusion for accurate victim localization