Periodic Reporting for period 1 - SoliDAIR (Solid, rapid and efficient adoption of Data, AI & Robotics applications in production)
Reporting period: 2023-10-01 to 2025-03-31
SoliDAIR brings together 10 manufacturing, technological and research experts to accelerate the uptake of Artificial Intelligence (AI) and Robotics in European manufacturing, using Data as an enabler. It will co-develop and demonstrate tailored solutions to digitalise and automate visual inspection and physical testing, enable predictive quality control and process optimisation. AI & Robotics systems are not extensively used in the production industry, because it is not clear whether they are safe and when or why they will fail. The SoliDAIR project tackles this problem by researching, developing and testing methods that are as solid and trustworthy as possible to be adopted by the European industry, while being cost-efficient to develop and replicate. New methods and tools will be developed by research and technology providers, which leverage the current state of the art in visual AI, AI for process data, and smart & collaborative Robotics. The developed technologies will be applied and demonstrated in 4 industry use cases to prove their functionality and applicability in real production environments. The objective is to improve production processes through digitalised and automated quality control for high volume, high rate and flexible manufacturing. The use cases are led by mature European manufacturing companies and represent problems that every producing industry struggles with. Solving them increases the process efficiency and flexibility, while improving the working conditions of the process staff. The developed methods shall be efficiently and easily adaptable and replicable, so they can be easily applied to new use cases outside the consortium. The accelerated adoption of the AI & Robotics solutions, tools and methods solutions will facilitate a strong increase in the competitiveness and sustainability of the EU manufacturing companies across sectors.
WP2 - Methodological framework & Generic modules: Task 2.2 "Visual AI theme", 2.3 "Data & AI theme: AI Based on Sensor Data" and 2.4 "Robotics & AI theme: Collaborative robots enhanced with AI" have been finished with a strong focus on generalizability and transferability. The results are used for the implementation in WP3 "Use case implementation and deployment". Task 2.1 "Generalised methodological framework" is currently paused and will be reactivated after the feedback from the implementation phase. The generic framework will be updated based on these lessons learned.
WP3 - Use case implementation and deployment: The activities and results for all the tasks (3.1 "Requirement definition and Data strategy for each use case", 3.2 "Implementation of user-centric AI enabled automated visual quality control system in UC BROSE", 3.3 "Implementation of Robotics and AI-enabled visual inspection in UC CIE", 3.4 "Implementation of robust AI quality prediction in matured, high-rate and high-volume production in UC BOSCH", 3.5 "Implementation of predictive quality control of a multi-step assembly process UC AUT") in WP3 indicate that overall, the WP is advancing according to plan. Efforts have been made in all the tasks to address the challenges emerging from the implementation of the generalized modules defined in WP2 "Methodological framework & Generic modules". Although there have been some delays that affected the planned workflow, solutions were adopted on time and the work was resumed without further repercussions.
WP4 - Use case demonstration, evaluation, and optimisation: Task 4.5 has commenced as planned. Efforts have been made to develop a set of technoeconomic KPIs. This KPI set has been distributed to use case leaders and will be used to assess the system performance during the upcoming demonstrations.
WP3 - Use case implementation and deployment: The activities and results for all the tasks (3.1 "Requirement definition and Data strategy for each use case", 3.2 "Implementation of user-centric AI enabled automated visual quality control system in UC BROSE", 3.3 "Implementation of Robotics and AI-enabled visual inspection in UC CIE", 3.4 "Implementation of robust AI quality prediction in matured, high-rate and high-volume production in UC BOSCH", 3.5 "Implementation of predictive quality control of a multi-step assembly process UC AUT") in WP3 indicate that overall, the WP is advancing according to plan. Efforts have been made in all the tasks to address the challenges emerging from the implementation of the generalized modules defined in WP2 "Methodological framework & Generic modules". Although there have been some delays that affected the planned workflow, solutions were adopted on time and the work was resumed without further repercussions.
WP4 - Use case demonstration, evaluation, and optimisation: Task 4.5 has commenced as planned. Efforts have been made to develop a set of technoeconomic KPIs. This KPI set has been distributed to use case leaders and will be used to assess the system performance during the upcoming demonstrations.
A generalised methodological framework was developed. This includes amongst others Synthetic Data Generation (images and process data) for industrial applications. In this field the developed and implemented physical-inspired synthetic data generation framework (GenAI to add realism) can be highlighted. Further general developed modules are Hybrid Models, Reduced Order Models, High Fidelity Simulation and extended XAI.