Resources

Public deliverables 

D1.2 Legal and ethical requirements for human-machine interaction (PDF)

This report contains an overview of the integration of ethics into the first six months of the AI-PROFICIENT project (taking into account also an update from information retrieved after site visits done in November and December 2021). The overview includes a literature review of the current and past work upon Industrial AI ethics, a survey of related guides and principles, and observations regarding the state of the discipline and its particular character relative to other fields of AI ethics.

D1.3 Pilot-specific demonstration scenarios (PDF)

This report incorporates a specification of the demonstrator to be constructed per use case. This specification has been reviewed and includes a structured and unified approach to the specification of all use cases, with a common information regarding Gap Analysis, Stakeholders, Data Sources, Ethical issues, and a High Level design including Use Case and Sequence Diagrams, as well as a flowchart showing the expected contributions from different partners and the link to appropriate task activities within development work packages WP2-WP4.

D7.1 Roadmap for dissemination and communication first release (PDF)

This report sets out the dissemination and communication strategy as well as the plan to raise awareness, share knowledge, attract potential stakeholders in the context of the AI-PROFICIENT project, through various means, including the AI-PROFICIENT website, the use of Social Media, the distribution of communication material, publications in scientific and industrial journals, participation in events and organization of dedicated workshops with potential end-users and main outreach events. 

D7.2 Project identity kit and communication material (PDF)

This document will help you understand the essential elements of the AI-PROFICIENT identity. It explains how to use the identity and serves as a source of inspiration for you to continue building a strong brand people love to be a part of.

D1.5: AI-PROFICIENT system architecture

The aim of this document is to provide the architecture of the AI-PROFICIENT platform following the “ethics by design approach”.

D3.1: AI-PROFICIENT hybrid models and digital twins first version (state of the art, designing and specification)

This report presents the state of the art of hybrid modelling and digital twins, covering the techniques from first principles modelling to fully data based surrogate models and digital twins based on these approaches. On that basis, designing of the approach for selected use cases and the specifications in each case are described.

D3.3 System-level proactive maintenance strategy

This report is describing the AI-PROFICIENT service that aims at optimizing maintenance scheduling at system level.

D4.1: Human-machine interaction and feedback mechanisms (Design and specification)

This report incorporates the description of the end user’s requirements related to the use cases in terms of the feedback management and the description for each one to manage it for a reinforcement learning approach.

D4.4 AI-PROFICIENT approach for XA

The report covers current state-of-the-art analysis related to different XAI approaches – exploitation of semantic technologies, transparent machine learning models and post-hoc explainability techniques. Many of those have been utilized for the development of three XAI services explained in this report – surrogate explainable data-driven model, post-hoc explainable analysis module and auditability system. For all of those, the methodology, results, integration and application were explained.

D5.1: Communication middleware and IIoT interoperability – design and specification

The aim of this document is to provide the design and specification of AI-PROFICIENT platform middleware and interface towards smart components and edge AI.

D5.2 Semantic data model for integrated digital twins

This document provides an insight into the process how the sensor measurements and the sensor equipment are ontologically expressed. It documents the challenges and the considerations that were raised during this process. The data model is implemented in data layer for the AI-PROFICIENT AI-services. This integration is outlined in this deliverable

D6.1 AI-PROFICIENT validation methodology (preliminary version)

The aim of this document is to propose a generic methodology to measure the degree of compliance with different AI modules deployed in an industrial facility.

D6.6 AI-PROFICIENT validation methodology (final version)

This report is the final version of deliverable D6.1, and the aim is to provide a complete procedure and ways to validate the initially proposed methodology. This methodology intends to establish objective measurement criteria for the results obtained in the AI-Proficient project. These criteria will enable measuring the results obtained in the different use cases.

D2.3 Predictive AI analytics for component self-diagnostics (PDF)

This deliverable presents the advances made in the context of Work Package 2 Smart components and local AI at system edge that are related to the development of edge systems used for diagnosticate the assets in which they are embedded or run.

D2.4 Local AI for proactive maintenance support (PDF)

This deliverable presents the advances made in the context of Work Package 2 Smart components and local AI at system edge that are related to the development of edge systems used for the prognostics of the assets health in which they are embedded or run

D2.6 Smart components and local AI at system edge (PDF)

This deliverable is a public report that summarizes the activities and achievements reached at Work Package 2 within AI-PROFICIENT H2020 project. This WP has attempted to introduce local AI technologies at the system edge under the ‘smart component’ concept, close to the production lines, considering MEMS and PLCs information outputs. This differentiates from other technologies expected to work at ‘cloud’ level on most scenarios (and developed in other WPs).

D3.2 Predictive AI for process quality assurance (PDF)

This deliverable provides a brief description of how AI technologies are a relevant aspect to ensure the quality of the process and consequently the quality of the product in manufacturing processes. However, the deliverable is mainly focused on describing the algorithms for AI predictive analytics for specific use cases. This analysis has been based on the combination of specific knowledge of the processes, the historical data gathered from the monitoring of the processes and the use of ML algorithms.

D3.5 Future scenario-based decision making (PDF)

This deliverable is a public document of AI-PROFICIENT project delivered in the context of Task 3.5: Future scenario based decision-making and lifelong self-learning as a part of Work Package 3 Platform AI analytics and decision-making support, regarding the description of the AI-PROFICIENT service that aim providing life-long self/reinforcement-learning capabilities.

D5.4 Integration with AI4EU’s AI on-demand platform (PDF)

The deliverable describes AI-PROFICIENT’s actions in terms of Task 5.4 to align activities with the AI4EU platform. The Deliverable also presents the integration with AI-on-Demand platform to i) profit from software modules already available in the platform to download and integrate into AI-PROFICIENT, ii) contribute to the AI-on-Demand platform by submitting AI-PROFICIENT’s results to the platform and make them available for future initiatives to build upon and enhance.

D5.8 System integration and deployment (PDF)

The aim of this document is to provide the description of the activities in relation to system integration and deployment of AI-PROFICIENT platform. More specifically, it provides the summary of the main activities that have been performed in WP5.

Material

AI-PROFICIENT Brochure (version 1)

AI-PROFICIENT factsheet

AI-PROFICIENT Brochure (Version-2)

AI-PROFICIENT factsheet(Version-2)

AI-PROFICIENT Digital info pack

Infographics

Publications

  • Chaoub, A., Voisin, A., Cerisara, C., & Iung, B. (2021). Learning Representations with End-to-End Models for Improved Remaining Useful Life Prognostic. PHM Society European Conference, 6(1), 8. https://doi.org/10.36001/phme.2021.v6i1.2785
    Download

  • López de Calle – Etxabe, K., Gómez – Omella, M., & Garate – Perez, E. (2021). Divide, Propagate and Conquer: Splitting a Complex Diagnosis Problem for Early Detection of Faults in a Manufacturing Production Line. PHM Society European Conference, 6(1), 9. https://doi.org/10.36001/phme.2021.v6i1.3039
    Download
  • Esnaola-Gonzalez, Iker. (2021). Can Ontologies help making Machine Learning Systems Accountable?. 10.13140/RG.2.2.29339.18722.
    Download
  • Nguyen, Van-Thai & DO, Phuc & Voisin, Alexandre & Iung, Benoît. (2021). Reinforcement Learning for Maintenance Decision-Making of Multi-State Component Systems with Imperfect Maintenance. 2142-2149. 10.3850/978-981-18-2016-8_304-cd.
    Download
  • L. Berbakov and N. Tomašević, “Internet of Things Platform Architecture for Smart Factories,” 2021 International Balkan Conference on Communications and Networking (BalkanCom), 2021, pp. 157-160, doi: 10.1109/BalkanCom53780.2021.9593239.
  • López de Calle – Etxabe Kerman, Garate – Perez Eider, Arnaiz Aitor, “Towards a Circular Rotating Blade Wear Assessment Digital Twin for Manufacturing Lines”, IFAC-PapersOnLine, Volume 55, Issue 2, 2022, Pages 561-566, ISSN 2405-8963, https://doi.org/10.1016/j.ifacol.2022.04.253.
    Download
  • Anderson, Marc, and Karën Fort. (2022) 2022. “Human Where? A New Scale Defining Human Involvement in Technology Communities from an Ethical Standpoint”. The International Review of Information Ethics 31 (1). Edmonton, Canada. https://doi.org/10.29173/irie477.
    Download

  • Van-Thai Nguyen, Phuc Do, Alexandre Vosin, Benoit Iung, Artificial-intelligence-based maintenance decision-making and optimization for multi-state component systems, Reliability Engineering & System Safety, Volume 228, 2022, 108757, ISSN 0951-8320, https://doi.org/10.1016/j.ress.2022.108757.

  • Alaaeddine Chaoub, Christophe Cerisara, Alexandre Voisin, Benoît Iung. Towards interpreting deep learning models for industry 4.0 with gated mixture of experts. 30th European Signal Processing Conference, EUSIPCO 2022, Aug 2022, Belgrade, Serbia. (hal-03785546)
    Download

  • Anderson, M.M., Fort, K. From the ground up: developing a practical ethical methodology for integrating AI into industry. AI & Soc (2022). https://doi.org/10.1007/s00146-022-01531-x.

  • Marc M. Anderson, Some Ethical Reflections on the EU AI Act, IAIL 2022: 1st International Workshop on Imagining the AI Landscape After the AI Act, June 13, 2022, Amsterdam, Netherlands, CEUR Workshop Proceedings (CEUR-WS.org, ISSN 1613-0073), Vol-3221
    Download

  • Anderson, M.M., Fort, K. (2023). Ethical Internal Logistics 4.0: Observations and Suggestions from a Working Internal Logistics Case. In: Borangiu, T., Trentesaux, D., Leitão, P. (eds) Service Oriented, Holonic and Multi-Agent Manufacturing Systems for Industry of the Future. SOHOMA 2022. Studies in Computational Intelligence, vol 1083. Springer, Cham. https://doi.org/10.1007/978-3-031-24291-5_25

    Contact Us






      Newsletter






      Quiz

      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      How does Machine Learning relate to AI?

      Answer
      Machine Learning is a subset of AI
      Machine learning is a subset of AI. The idea is that machines will “learn” and improve at tasks over time rather than having humans continually input parameters. Machine learning is a practical application of AI.
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      What is Neural Networks?

      Answer
      Is a method in artificial intelligence that teaches computers to process data in a way that is inspired by the human brain​
      Neural networks are a class of machine learning algorithms. The neuron part of the neural is the computational component, and the network part is how the neurons are connected. Neural networks pass data among themselves, gathering more and more meaning as the data moves along. Because the networks are interconnected, more complex data can be processed more efficiently.​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      What is Artificial intelligence?​

      Answer
      Making a Machine intelligent​
      Artificial intelligence (AI) is intelligence—perceiving, synthesizing, and inferring information—demonstrated by machines, as opposed to intelligence displayed by non-human animals and humans​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      The characteristics of the computer system capable of thinking, reasoning and learning is known as

      Answer
      artificial intelligence​
      Artificial intelligence (AI) is the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. The term is frequently applied to the project of developing systems endowed with the intellectual processes characteristic of humans, such as the ability to reason, discover meaning, generalize, or learn from past experience.​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      The first AI programming language was called:​

      Answer
      IPL
      Information Processing Language (IPL) is a programming language created by Allen Newell, Cliff Shaw, and Herbert A. Simon at RAND Corporation and the Carnegie Institute of Technology about 1956. This is the first AI programming Language known.​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      An AI technique that allows computers to understand associations and relationships between objects and events is called:​

      Answer
      pattern matching​
      Pattern matching is a technique used in AI which allows a computer program to identify a specific pattern that is present in a given data set and take appropriate actions based on the results of the analysis.​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      Artificial Intelligence has its expansion in the following application.

      Answer
      All of the above​
      From manufacturing to marketing, businesses across multiple industries are leveraging artificial intelligence to optimize their processes and services.
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      Who is the father of artificial intelligence?​

      Answer
      John McCarthy​
      John McCarthy (September 4, 1927 – October 24, 2011) was an American computer scientist and cognitive scientist. He was one of the founders of the discipline of artificial intelligence.​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      When did the term "artificial intelligence" first appear?​

      Answer
      In 1956​
      In 1956, at a conference at Dartmouth College, in Hanover, New Hampshire, the term "artificial intelligence" was coined​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      Which is not the commonly used programming language for AI?​

      Answer
      Perl​
      Perl is not the commonly used programming language for AI. ​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      Great, now that you understand the basics of Artificial Intelligence, we suggest that you glance at our project documents highlighting the improvements that AI services introduced in the manufacturing industry.
      Within the AI-PROFICIENT project, we piloted seven different use cases run across different manufacturing sites of INEOS and Continental. Our solutions aim at improving manufacturing efficiency while keeping humans in control.
      The following questions are based on the outcomes of AI-PROFICIENT use cases and should give you an idea of the practical advantages of AI for manufacturing. Click next to get started.
      Answer
      Great, now that you understand the basics of Artificial Intelligence, we suggest that you glance at our project documents highlighting the improvements that AI services introduced in the manufacturing industry.
      Within the AI-PROFICIENT project, we piloted seven different use cases run across different manufacturing sites of INEOS and Continental. Our solutions aim at improving manufacturing efficiency while keeping humans in control.
      The following questions are based on the outcomes of AI-PROFICIENT use cases and should give you an idea of the practical advantages of AI for manufacturing. Click next to get started.
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      By how much percent AI implementation decreases raw material waste in tyres production?​

      Answer
      8%
      Combining Explainable AI algorithms with the global optimisation algorithms in the rubber extrusion allows continuous suggestion of the operator's optimal machinery parameters. This reduces the time when the rubber extrusion process is set and thus reduces the time when the suboptimal rubber outcome that cannot be used in further tire production.
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      Which of the following is a polymer extrusion process:​

      Answer
      A chemico-mechanical process​
      Chemically, the process involves the melting and flow of a polymer material under heat, which can result in chemical reactions such as degradation or cross-linking of the polymer chains. The chemical properties of the polymer material and the extrusion conditions, such as temperature, pressure, and residence time, can affect the final properties of the extruded product.​

      Mechanically, the process involves the deformation of the molten polymer material under pressure as it is forced through a die orifice to form a continuous shape. The mechanical properties of the polymer material, such as its elasticity, viscosity, and melt strength, determine its ability to withstand the stresses and strains imposed during the extrusion process.​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      Which are the key elements of an optimization problem:​

      Answer
      The objective function, the decision variables and the constraints​
      The objective function defines the goal of the optimization problem, whether it is to maximize or minimize a certain value. The decision variables are the variables that can be adjusted or controlled to achieve the optimal solution to the objective function. The constraints are the limitations or restrictions that must be satisfied during the optimization process.​

      The other two options mentioned in the question are not related to the key elements of an optimization problem. Hyperparameters, validation schemes, and accuracy metrics are more related to machine learning or data science problems. The differential hypothesis is not a common term used in optimization problems. ​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      When the blade of the cutting system starts to wear out but it can keep cutting:​

      Answer
      It might work well but chances are eventually it will compromise the quality of the final product.​
      As the blade of the cutting system starts to wear out, it will gradually lose its sharpness and efficiency, which can affect the quality of the cut and increase the cutting forces. If the blade is allowed to continue cutting even after it has worn out, it can lead to several negative consequences.
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      According to the literature, which are the three main sources of variation when trying to model a cutting process in an industrial setup:​

      Answer
      Workpiece factor, feed factors and saw blade factors​
      In a well-controlled manufacturing environment, the cutting process should be designed to be robust and less sensitive to operator variation. Regarding to the environmental factors, they are less significant compared to other sources of variation.
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      Developing the AI service which will improve tread alignment in part of the Combiline depends upon marking pictures of the tread pieces, which will then be used to train the AI. How should the developers and industrial partner approach this picture marking task for AI training?​

      Answer
      The developers should clearly specify how many pictures need to be taken, how long the process will take, and discuss with the industrial partner who will do it.​
      An ethical approach to this task involves clearly specifying how many pictures, and how much added work time it involves. On that basis the developer can discuss with the industrial partner who should do the task, e.g. whether the developers themselves are better placed to do it, or whether the industrial partner’s operators should do it. If the latter, it should be made clear to the operators whether they can refuse to do it, or it is an obligation which makes up part of their work. ​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      What is explainable AI and how it can help in manufacturing?​

      Answer
      XAI stands for "explainable artificial intelligence" and it can help in manufacturing by improving process control, enabling predictive maintenance, enhancing quality control, and improving safety.​
      XAI stands for explainable artificial intelligence and provides insights into decision-making processes, identifies patterns and anomalies in equipment data, and detects defects in real-time. This allows manufacturers to make better decisions, reduce downtime and maintenance costs, and address problems quickly.​
      General
      1
      2
      3
      4
      5
      6
      7
      8
      9
      10
      UC
      11
      12
      13
      14
      15
      16
      17
      18

      How should the developers engage the operators when developing the AI service and related apps and HMIs, with the goal of simplifying the job of confirming correct chemical product bag placement before production? ​

      Answer
      The developers should work closely with the operators to design the services, choose the HMIs and adapt the apps, according to the actual work context. ​
      To implement an ethical by design user centered approach, the developers should work closely with the operators who will actually use the service. They should take care to understand the specific work conditions (dust, noise, time for tasks, etc.), discuss the planned solutions with the operator users, design initial mockups and discuss these with the operator users regularly in an iterative process as they develop the service. ​