Page 532 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

(continued)

Item Details
Model Train- Personalization involves data analysis and insights generation and it follows
ing and two different approaches with the aim of 1) recommending optimal levels
Fine-Tuning of automation and 2) predicting the expected level of productivity, thereby
improving operational efficiency and identifying potential areas for improve-
ment. Firstly, a statistical analysis are performed on the integrated data
to develop a data-driven rule-based approach, augmented with expert
knowledge, resulting in a set of hard-coded rules. Additionally, machine
learning techniques are employed to identify unexpected characteristics
influencing productivity, utilizing explainable models like decision trees
and more complex models such as Support Vector Machines and Artificial
Neural Networks. Model-agnostic methods like SHapley Additive exPlana-
tions (SHAP) and Local Interpretable Model-agnostic Explanations (LIME) are
used to interpret these complex models. Finally, an easy-to-use Augmented
Reality (AR) add-on is developed, leveraging insights from rule-based and
machine-learning approaches. This add-on is provided following a container-
ized microservice approach for seamless integration and deployment.
Testbeds or Advanced Manufacturing Technologies (AMT) and the Deep Tech approach,
Pilot Deploy- converging digital and physical realms, are crucial. Industrial operators need
ments new emotional and relational skills for interacting with intelligent machines,
while training designers require a holistic, multidisciplinary approach.
Extended Reality and the Industrial Metaverse (XR/IM) are key to innovating
production and fostering a skilled, resilient, and creative workforce, essential
for successful industrial transitions and competitiveness. Industry training for
automation, specifically programming a CNC machines and collaborative
robot with hands-on exercises. This experience-based method accelerates
learning compared to traditional methods.

Code reposi- Not available
tories

2 Use Case Description

2�1 Description

• The project aims to define, support and accelerate the adoption of adaptive distributed
XR Human-Machine Interaction (HMI) in the industry, by introducing to the market a
modular and distributed XR Toolkit for enhancing human-centered manufacturing based
on adaptive, personalized and distributed XR industrial machine interfaces for a safer,
more inclusive and well-integrated working environment.
• XTENDIT integrates XR, Internet of Things (IoT), AI and mobile devices for promoting a
more adaptive and informed use of machine automation to help industries and machine
operators. Specifically, XTENDIT is a XR Toolkit supporting human decision-making and
thus well-being, safety and productivity of the workforce in manufacturing settings. As an
example of the intelligence augmentation paradigm, that sees human and AI working
together as a symbiotic system, XTENDIT proposes AI to augment human users’ decision-
making process and capabilities by providing otherwise hidden or inaccessible data-
driven insights.

XTENDIT use case’s objectives are:

• 1_ To implement an interaction design model for Human-Centered Automation and
AI-Human Augmentation, based on personalized, adaptive, accessible, distributed and
context-sensitive user interfaces;

496