ECSEL JU Project AI4DI

AI for Digital Industry

Summary description of the project objectives

The mission of the AI4DI project is to devise a harmonized pan-European AI framework for the Manufacturing and Process technology, including prove of relevance of industrial manufacturing and process applications, which may become the optimal resilience for Manufacturing and Process technology. A plausible and reproducible approach for a wide implementation of AI methods in manufacturing industry is chosen by means of the AI deployment plan. The basic structure comprises 7 key targets, which are used in eight Work Packages, to contribute to 5 Objectives.

The key targets (KT) group the specialist environment required for digitizing manufacturing processes according the criteria area of expertise and necessary functionality. Each KT represents a field of activity and the corresponding target at the same time. In any case, actions are taken within the KT to tackle the challenges that currently prevent achieving the 5 objectives. For example, KT1 addresses all actions concerning the control and optimization of heterogeneous systems (e.g. batch processes) with simultaneously targeting an evolvement of a common AI method understanding in this context. It is intended that each Use Case makes use of several KT since these are intended to interact by sharing data and expertise. The precise implementation of the planned actions within the KT is described in the Work Packages.

Description of the work to be performed

In order to achieve the ambitious objectives, AI4DI focusses on four pillars that interact with each other such that they generate a) the methods, b) an AI Deployment plan, and c) the quality reference AI Metrics required for the verification, validation, and AI in real industrial life of automated systems. The advantage of such 4-pillar structure, is that it uses techniques from different expertise areas, which support a 4-dimensional understanding of such a complex problem, being an optimal solution, which can lead to a harmonized certification of Manufacturing and Process technology. Using this approach (covered by reference partners out of this domains), will lead to the alignment with various stakeholders to define a common strategy for virtual and physical relevance and applicability to homologate Digital Industry Industrial processes.

Expected final results and demonstrators

The 7 Use Cases are intended to validate the 7 KTs and contribute to the achievement of the 5 Objectives.

  1. Digital ECS value chain for virtual factory, uses AI methods as distributed intelligence, which is scheduling slots in a manufacturing process of vehicles and also orders components from suppliers who are also using AI affecting their production process to produce and deliver the components in time.
  2. Digital Twin Factory, aims at a Digital Twin for the AI-Based Optimization of Heterogeneous Manufacturing Processes.
  3. AI-based FMEA generator, sets up an AI based Knowledge base that combines the knowledge of existing Technology FMEAs.
  4. Safe human interaction with robotics machinery and tools, addresses the deployment of AI-based Human/Machine Interfaces striving the improvement of the working conditions and productivity of human operators interacting with complex stationary machines.
  5. Change detection applications, is composed of several applications addressing the challenge of change detection in real-time sensor data using AI methods. The resulting information will be used in order to take actions to efficiently operate systems, e.g. cyber-physical systems, machines or processes.
  6. Robotics application, is also composed of several applications addressing mobile or rather industrial robots that offer manifold manipulating tasks. AI methods are adapted in order to control the complexity associated with this flexibility as well as to improve the human machine collaboration.
  7. AI implementation, groups the AI4DI activities on implementing AI methods on the wide range of available resources (IoT/edge/Cloud) e.g. for the purpose of an overall traffic management based on data from individual vehicles (bus or car).

Potential AI4DI Impact (including the socio-economic impact and the wider societal implications of the project)

The ongoing revolution in industrial production – Industry 4.0 – results from a confluence of fast-developing technologies. These range from a variety of digital technologies (such as 3D printing, the Internet of Things, advanced robotics) and new materials (bio- and nano-based) to new processes (for example, data-driven production, artificial intelligence and synthetic biology). Europe possesses considerable strengths, and in some cases global leadership, in a number of these technologies. This is particularly true of artificial intelligence, digital security and connectivity.

The current AI industry had been built around a centralized distribution paradigm where machine learning solutions are delivered as a part of cloud-based APIs and software packages deployed on remote servers of AI providers. The future requires a paradigm shift by moving toward decentralized AI that can run and train at the edge on local intelligent devices in industrial applications or make decisions in decentralized networks like blockchain.

The transition to decentralized AI is enabled by new technologies, that allow crowd-training of ML algorithms, device-centred AI that runs and trains ML models on mobile devices, and the use of AI in decentralized autonomous organizations on blockchain networks.

Intelligence on an edge device gives it the ability to process information locally and respond quickly to situations, instead of communicating with a central cloud or server. For instance, an autonomous vehicle must respond in real-time to what’s happening on the road. Decisions are time-sensitive and latency could prove fatal. The same requirements are in several manufacturing processes. A goal and outcome of the project is definitely to change the mindset of the public, which is still hesitant and sometimes anxious in respect to new technologies, and to open the public opinion to the idea and the power of the possibilities with AI-driven technologies.

With the dawn of artificial intelligence, many new jobs will be created, but some traditional ones will disappear and most will be transformed. To meet this social challenge, AI-specific expertise needs to be formed by teaching and training, current curriculums in European schools and at universities need to be revised and updated. AI talents in Europe need to be developed and fostered by setting up dedicated training schemes: digital skills, competencies in science, technology, engineering and mathematics (STEM), entrepreneurship and creativity need to be supported. A huge change is coming to societies, and it´s a major task to inspire and fascinate a broad part of the society of positive effects of AI: Not just regarding technologies, but tangible topics also lying behind like healthcare, education and environmental protection.

As with any transformative technology, artificial intelligence may raise new ethical and legal questions, related to liability or potentially biased decision-making. New technologies should not mean new values. To help progressing ethical guidelines on AI development, it is necessary to bring together all relevant stakeholders in a European AI Alliance. AI4DI will actively support any activities guarded by the Commission, also in means of guidance on the interpretation of the Product Liability Directive in the light of technological developments, to ensure legal clarity for consumers and producers in case of defective products.

ECS Strategic Research Agenda focus areas:


ECSEL Call 2018

Start date:  05/2019

Duration: 36 months

Project coordinator:

Reiner John
reiner.john@infineon.com

Infineon Technologies

Germany

Number of partners: 41

Number of countries: 12

Total investment: M€ 30

More on: