Short title: A2P (Assist to Produce)

Long title: A Framework for Adaptive Worker Assistance Systems in the Industry

Abstract:

Adaptive worker assistance systems for industrial applications have been shown to increase productivity, decrease error rates, and reduce physical and cognitive load. However, despite the aforementioned benefits, the penetration of these systems in the industry has not progressed beyond application-specific prototypes. The reasons for the low penetration are lack of implementation standards, high deployment efforts caused by time-consuming analysis and costly customization, inefficient human-machine interaction, and the need for privacy-preserving data management. A2P (Assist to Produce) aims to address the challenges adaptive worker assistance systems face during the development and deployment phase. Using human activity recognition, the project aspires to create a unified framework for adaptive cognitive and physical worker assistance systems. Furthermore public datasets for the recognition of industrial activities and identification of work steps with further need of support will be created. Lastly, the project will set a baseline for sustainable privacy preservation in manufacturing by employing anonymisation techniques for private data and secure multi-party AI training with federated learning.

Duration: 01.05.2022. – 30.04.2025

Funding agency: Austrian Research Promotion Agency (FFG)

Partners: 

 

Contact:

Dipl.-Ing. David Kostolani

E-Mail: david.kostolani@tuwien.ac.at