IFAC World Congress - Human in the Loop of Artificial Intelligence in Smart Maintenance and Manufacturing Systems track
The Open Invited track - Human in the Loop of Artificial Intelligence in Smart Maintenance and Manufacturing Systems will be organised at the IFAC World Congress, the flagship event of the International Federation of Automatic Control, that takes place every 3 years.
The co-organisers of the track are:
- Jon Bokrantz, Chalmers University of Technology, Sweden
- Christos Emmanouilidis, University of Groningen, The Netherlands (corresponding organiser and STAR project partner)
- Paulo Leitao, Polytechnic Institute of Bragança, Portugal
- M.Rožanec, JSI/QLECTOR, Slovenia (STAR project partner)
- Thorsten Wuest, West Virginia University, USA
The Final extension for paper submissions is now set for 11 November 2022. The Congress will take place in Yokohama, Japan, 9-14 July 2023 and is endorsed by IFAC TC5.1 (Manufacturing Plant Control) and AMEST WG (Advanced Maintenance Engineering Services and Technology).
Session Abstract:
AI-driven human augmentation or industrial automation have seen many applications in maintenance and manufacturing. High expectations are set regarding AI-driven solutions and automated outcomes, but the role of the Human in the Loop in producing these outcomes is less well explored. This is surprising given that Human integration in Sociotechnical Systems has long been studied. Much is expected to be achieved in automated manner from Machine Learning in industrial systems, leaving the possibility for properly integrating human knowledge and human capabilities insufficiently exploited. Yet, the application practice of Machine Learning and broader AI in Maintenance and Manufacturing provides ample evidence of brittleness of derived solutions in the face of limited or new data, changing contexts, or evolving situations. AI communities seek to address such challenges with approaches such as Transfer Learning. More recently Active Learning has been explored to better focus on the integration of Human Interaction, and therefore the Human Physical and Cognitive Capabilities in the AI Loop. The interest in addressing Human – Centricity in Industry 5.0 often targets high conceptual, abstraction, and design levels, and does not sufficiently target the interactive and operational engagement of the Human in the AI Loop. This pattern is changing, especially in domains with high performance, safety or ethics requirements, with research targeting mixed or sliding autonomy and decision making, shared contexts and collaboration workspaces. Such approaches deserve further research in maintenance, as well as manufacturing shop floor contexts. The simplest cases are human – annotated data to drive machine learning. Modelling and integration of domain knowledge with data via knowledge graphs and ontologies is also pursued. Employing human operators, workers, or engineering staff as a source of observation, knowledge, decision, or action is another example. The collaboration of human and non-human (AI-driven) cooperating agents in industrial systems is a further step. Considering the above, the effective integration of humans in the Machine Learning and Broader AI Loop for Maintenance and Manufacturing is the focus of this track.