UBITECH participated at the kick-off meeting hosted by Ecole Nationale Superieure D’arts Et Metiers in Arts et Métiers in Aix-en-Provence (Paris, France), on January 19-20, 2023, of the MARS Research and Innovation Action, officially started on January 1st, 2023. The project is funded by European Commission under Horizon Europe Programme (Grant Agreement No. 101091783) and spans on the period January 2023 – December 2026. MARS aims to enable SMEs to access advanced European breakthrough innovations in the field of AI-driven digital manufacturing processes and enter into process chains that are geographically distributed.
Specifically, by gathering diverse expertise coming from complementary European partners, MARS will develop Industry 4.0 emerging technologies including digital twins of products, processes and machines, bio-intelligent production devices with local intelligence and high sensing coverage, central intelligence with fleet learning approaches, data-driven manufacturing process models from different sources, blockchain technology for data hashing, traceability and securitization, multi-agent based manufacturing planning, multi-criteria intelligent optimization of processes and resources especially addressing environmental footprint.
As a result, the impact of the project will lie into introducing radical flexibility in all different aspects of manufacturing processes, in particular by redefining the process route, raw material, resources, technology, throughput, manufacturing site, delivery date in no time, while keeping up with product’s requirements, proven product quality and sustainability of both processes and products. By demonstrating its results on two case studies exhibiting advanced manufacturing processes (incl. homogeneous and heterogenous data), MARS will show how SMEs can decrease time delivery under difficult economical boundary conditions, while targeting ambitious energy-saving environmental objectives.
Within MARS, UBITECH defines a common semantic model to encapsulate the main concepts of the manufacturing process, as well as a reference architecture for federated learning in distributed manufacturing. UBITECH’s activities leverage capabilities of the federated learning paradigm in a way that will facilitate rapid deployment of new AI enabled solutions for zero defect manufacturing. Locally trained AI models will be aggregated into a single model via the federated learning algorithms; this model will then be propagated back to existing nodes.
The common models will serve both aspects of process planning: a) Optimisation: For optimisation, the model will take into account both hard and soft criteria and output an evaluation of the stability of the process. b) Control: For process control the model will offer decision support by providing corrections to the parameters which, while still allowing the process to remain in a stable state, will offer better behaviour in terms of cost and use of resources.
