Go to JKU Homepage
Institute of Pervasive Computing
What's that?

Institutes, schools, other departments, and programs create their own web content and menus.

To help you better navigate the site, see here where you are at the moment.


Reducing Energy and Waste using AI.

Funding FFG, Call: AI for Green 2021 (KP)    
Project no. FO999892233    
Duration 2022-2025
Consortium Pro2Future GmbH*, Johannes Kepler Universität Linz, Technische Universität Graz, Lenzing AG
Role Project partner

The textile and fashion industry are responsible for 10% of global greenhouse gas emissions and 20% of global land and water pollution. While renewable and sustainable textile fibre, in particular lyocell and viscose, choices exist, they are non-dominant on the market. Part of the reason is that production costs especially in Europe are high and increasing, energy costs already rose and will continue to rise, especially when more of society switches to battery powered transportation. Thus, optimisation of reducing production resource usage, like energy, is the main goal of this project, to increase competitiveness of these sustainable fibres.

AI Powered Tools and Methods have been showcased in various domains and are seen as an enabler for a sustainable future as envisioned by the UN sustainable development goals. High impact AI research in production environments is typically hindered by lack of data or funding. With company partner Lenzing, a world-leading manufacturer of wood-based cellulose fibres, and their highly instrumented multi-step production process with history data of 20 years, a perfect testbed for applying AI for sustainability in manufacturing and material goods industry is provided. Their production line is one out of several others which could benefit from AI based optimisation.

Directly measurable optimisation benefit would be the reduction of energy, waste and storage capacity required by an internal upcycling process, by avoiding the production of inferior quality goods. The nature of the production process hides causal relationships between sensor values and quality outcomes and not all quality attributes can be measures all the time creating causal and data gaps throughout the production line. These gaps, we want to close by applying causality methods and deep learning-based quality predictions.

Another problem we want to tackle is that AI in manufacturing is hindered also by two  major human factors: (i) threat of replacement of human workforce and (ii) lack of trust towards AI based decision making. Thus, we aim to build interfaces where human intelligence is augmented by machine intelligence using explainable AI techniques. These interfaces should enable human workers to be in control and provide human oversight and thus create the necessary trust for (ii) and mitigate (i). In addition, we aim to provide interfaces for what-if analysis so human operators or managers can investigate also what would be necessary to accomplish another desired outcome of the overall production line. This again would accomplish the enabling of trust in AI systems for production. Finally, we want to investigate how our findings can be applied to other production domains and influence the reduction of energy and waste through AI (REWAI) towards a more sustainable future.