Technologies such as Artificial Intelligence (AI) and eXtended Reality (XR) are having – and increasingly will have – a strong impact on Italy's digitalization process and its competitiveness’ growth, including in the manufacturing sector.
Through the expansion of XR WEAVR platform and the integration of new AI-based and human-AI teaming modules, TXT intends to pave the way towards a new paradigm of Digital Manufacturing, supporting industrial partners throughout the entire life cycle of the production process, while also enabling the scalability of the traditional approach to the development of factory procedures, based on the digitization of manuals and subsequent validation and use in the field.
Co-funded by the Ministry of Enterprises and Made in Italy, through the Innovation Agreements (with PNRR funds), the project SOFIA - SOFtware platform for explainable and ethical Artificial Intelligence for the worker 4.0 aims to improve the WEAVR product (the platform for building, deploying, executing and monitoring AR/VR applications owned by TXT) through the development of new modules based on Artificial Intelligence technologies and on the optimization of human-machine/AI interactions.
It will therefore be possible to create a set of explainable and ethical AI services centered on the factory worker, facilitating the re-location of the human in the modern factory and its collaboration with AI, overcoming the barriers in the adoption of Augmented Reality and Artificial Intelligence in supporting factory operations.
Fundamental enabling technology to which the project is aimed: Artificial Intelligence.
Area of intervention in which the technology to be developed has an impact: Artificial Intelligence and Robotics.
General guidelines developed in the R&D project:
1. Enabling AI technologies, such as intuitive AI, ethical AI, human-controlled AI, unsupervised machine learning, data efficiency, and advanced human-machine and machine-to-machine interactions;
2. Human-centered AI-related technologies for AI-based solutions.
Drastically reduce the time to digitize and optimize factory procedures from real experience in the field.
This is achieved through the following modules:
1. Unsupervised learning module that observes what is happening in the factory, learns the typical behaviors of the assets under observation, and learns the activities that led to positive results, those that did not have an effect, and those that had negative effects.
2. A module for predicting behaviors and prescribing procedures that, through the use of machine learning algorithms, captures real situations in the real field by predicting the future behaviors of the asset and - based on Explainable Artificial Intelligence (eXplainable AI) algorithms - defines the most likely optimized procedure and possible alternatives.
To build an approach with humans at the center, based on a module of explainable and augmented interfaces.
Through a set of interfaces and interaction patterns, connected to the explainable AI, the key elements are made available to the worker to understand the motivations of the decisions suggested by the AI. This "glass-box" approach enables the execution of collaborative workflows between humans and AI.
Carry out the study, design, and implementation of effective collaborative activities between worker and explainable AI, based on two concrete use cases that will be implemented and evaluated by the end-user's staff.
The test will be carried out both in a simulated environment in Virtual Reality, using the digital twins of the assets, and then in Augmented Reality in the field:
• Assembly/disassembly of compressors made available by SIAD Macchine Impianti;
• Predictive maintenance on CNH Industrial agricultural machinery systems.
CUP ID: B29J23000130005 (TXT) - B19J23000120005 (SIAD) - B89J23000840005 (DBLUE) - B19J23000130005 (CNHi) - B49J23000270005 (POLIMI)
Project duration: 36 months
Start date: 1 April 2023
End date: 31 March 2026
Funded by: Ministery of Enterprises and in Italy (Innovation Agreements – PNRR)
Partnership: TXT e-tech, CNH Industrial, SIAD Macchine e Impianti, Deep Blue, Politecnico di Milano
Project Budget: € 4.954.937,50 (out of which € 2.521.951,44 as PNRR contribution)