During the second reporting period significant progress toward the achievement of the objectives (O1-O6) were made:
O1 - A functional prototype of the Smart Voting-Wristband Arrowhead application device has been designed and manufactured, integrating a Bosch BMI270 motion sensor for accurate thumb gesture detection. The system architecture has been built using the Eclipse Arrowhead Framework, with secure communication established between the wristbands and a central server. A control panel has been developed using the Python Streamlit library for easy creation and configuration of voting sessions and real-time voting results.
O2 - There is a significant advancement of the AI toolbox. The toolbox now includes some well-defined functionality regarding design templates, examples, and reusable tools, describes best practices, that help the designers to enhance reliability, includes prompt evaluation to test industrial proof-of-concepts. The algorithms, tools and methods are identified for the use cases, information is collected about them containing their descriptions, plans and connections to UCs. They show solid foundation on the SOTA and significant advancement beyond.
O3 - The definition of Eclipse Arrowhead v5.0 and emerging implementation of v5.0 in both Java and Golang with new documentation partly based on the Arrowhead DSL provides an easier to install and use AI for automation integration technology. Particularly the implementation of security is simplified compared to v4.6 of Eclipse Arrowhead. Considerable contributions to other fundamentals like composability, real time and deployment has been made. Many not directly visible to most users. Most of these contributions provides improved robustness, usability and engineering efficiency.
O4 - WP4 has taken the review feedback into consideration to include more of the project use case partners involvement into the digital reference model and OAP. The simulation-based optimization approach for production planning of a single wafer fab has been designed and initially tested. Comparison of the planning UC with the concept of the ecological operating curve from UC 18 as well as the sustainability lobe of the Digital Reference was studied. Current structures in the Digital Reference Models regarding the battery management system of an electric vehicle were evaluated, suggested extensions for a product carbon footprint have been derived.
O5 - Various advancements were achieved across multiple facets of autonomous systems and AI-driven technologies, demonstrating the potential for transformative impacts in industrial settings. The continuous iterations of requirements engineering, application, and evaluation of developed solutions in lab environments and real production use cases ensured robust validation and refinement of AI techniques across different industrial domains. Intensive testing, validation and the influence on sustainability KPIs will be shown in the subsequent reporting period. For the majority of use cases an ontology was worked out in the second year. Many results of the use cases have a very positive influence on the working conditions (objective 6).
O6 - A self-assessment tool was designed, which is a chatbot built using a Retrieval-Augmented Generation (RAG) architecture. The chatbot helps users assess their AI systems against the EU AI Act by offering personalized compliance support through an intuitive conversational interface. The Quantitative Effect and Technology Acceptance Model (QETAM) was used to test the user acceptance of this tool.
