Testing and Experimentation Facility for Edge AI

Topic description

ExpectedOutcome:

Outcomes and deliverables:

• Wide availability of world-class experimentation facilities in Europe for edge AI microelectronic components based on advanced low-power computing technologies;
• Service provisioning to a wide range of users: from semiconductor manufacturers and fabless companies, to end-user industries, SMEs and start-ups, requiring a varying degree of support in their further market development;
• Fully tested and validated European IP and edge AI products based on European technology, ready for market deployment.

The overall aim of the Testing and Experimentation Facility (TEF) for Edge AI hardware is to ensure the availability in Europe of trusted, high-performance, low-power edge components and technologies to support the massive data-processing requirements of AI and the digital transformation. The emergence of novel computing paradigms, such as neuromorphic, provides an opportunity for the European industry to take the lead in a new generation of edge and distributed computing systems.

The Edge AI TEF will also contribute to societal challenges, in terms of:

• supporting the implementation of the Green Deal, by reducing energy consumption versus centralised computing solutions, also thanks to ground-breaking ultra-low power operation (100 to 1000 times more efficient) and very limited use of the data infrastructure (data transmission consumes even more energy than computing); further, edge AI components enable environmental monitoring as well as increased energy-efficiency of buildings and smart cities;
• protection of personal data and privacy by processing within the edge device;
• security and resilience avoiding or limiting dependency on the network;
• safety and efficiency by reducing latency to near-real-time (e.g. for automotive or manufacturing applications).
• European sovereignty in generation and management of sensitive data with trusted, secure components;
• Open technological autonomy by reducing and diversifying strategic dependencies on components that may expose the EU to supply chain disruptions and security vulnerabilities, or not respecting European values in terms of safety, privacy, AI, equality, ethics or environmental aspects.

This project aims at delivering a European platform that will enable companies to develop, test and experiment AI product prototypes based on advanced low-power computing technologies, custom-designed for their application environment. It involves the set-up of facilities for design, fabrication, integration and verification, employing leading-edge equipment with advanced semiconductor process and fabrication technologies. Such a platform will be able to bring novel embedded AI technologies to high Technology Readiness Levels (TRLs), delivering prototypes for field validation that can significantly accelerate industrialization and time-to-market for edge AI components, resulting in a competitive advantage for the European ecosystem.

The TEF will realise a joint distributed platform (pilot line) to be led by EU centres of excellence in microelectronics technologies for advanced computing and key industrial actors of the microelectronics ecosystem such as integrated device manufacturers (IDMs) foundries and system integrators. The platform will be open to exploitation by EU industries, including SMEs, through direct support or through projects funded by other European or National programmes. In particular, strong cooperation is envisaged with the Joint Undertaking on Key Digital Technologies, which may support projects in close coordination with the TEF on Edge AI, as well as other EU partnerships in the area of AI, HPC, 5G/6G and cloud computing. Opportunities for future synergies with related actions on quantum computing infrastructure will also be evaluated.

Through the collaboration of a network of European Digital Innovation Hubs, the Edge AI TEF will provide also SMEs and Start-ups with fast and simple access to world-class testing facilities, rich networks of stakeholders and potential customers.

In cooperation with the sectorial TEFs in Manufacturing, Agri-food, Healthcare, Smart communities, the AI components delivered by the Edge AI TEF can be implemented in connected autonomous objects to be integrated in a distributed AI platform for applications such as industrial Internet of Things (IoT), precision agriculture, autonomous vehicles, implanted/wearable medical devices.

The selection of projects to be supported by the Edge AI TEF will be based on the character of innovation, on the business potential and on the strategic dimension for Europe.

The TEF infrastructure will put in place advanced semiconductor equipment that may also serve as a basis for quantum computing technologies, therefore opportunities for synergies with related Digital Europe actions on quantum computing infrastructure will be evaluated.

The first phase of the project will focus on the procurement, installation and preparation of the infrastructure, developing the necessary interfaces and testing of existing solutions, and the development of frameworks for hardware/software co-design and prototyping. In the following phases, the project will develop innovative prototypes employing the most advanced architectures and technologies beyond the state-of-the-art, incorporating energy efficiency, operational trust and security, which will be tested and validated in the respective field of application, before transfer for industrialization and production by IDMs or foundries.

The activities to be implemented based on this Work Programme are the following:

• Procurement and preparation of the infrastructure - equipment installation, design tool finalisation, definition of devices with users, distribution;
• Production of System Exploration Platform software for emulation of hardware;
• Progress of key building blocks (such as sensing, in-memory computing, 3D architectures) to higher TRLs and validation of their design kits; development of design and prototyping frameworks;
• Integration of building blocks into solutions, implementing rules and procedures for testing and transfer to industry.

Implementation: European Commission