Using AI to accelerate recycling

From the steel that goes into our cars to the electrical components inside refrigerators, producing the goods that keep our modern world running comes at a steep environmental price. “They are very carbon intensive to produce, and deplete mineral resources,” explains Naila Rana Andira, a Scientific Researcher at the OFFIS - Institute for Information Technology, whose work focuses on how manufacturing processes can be made more sustainable through information systems. 

Some of the worn-out components in our old appliances can be recycled, but this should be the last resort in a waste hierarchy that prioritises reuse and repair. “We look into whether parts are salvageable, whether they can still be repaired, and how much environmental impact could be reduced by repairing them rather than recycling,” says Naila.

However, the current process for assessing the quality and potential of these old parts is laborious and time-consuming, as workers need to manually check each one. This inefficiency has an environmental cost. “In manufacturing, productivity is tied to sustainability. If the manufacturing process takes longer than it should, it can lead to higher use of resources like electricity and water,” says Naila. 

Her team from OFFIS are exploring how AI can be harnessed to perform a first check of the viability of these used components, cutting down the time it takes. Naila has calculated that, despite the environmental footprint associated with training AI models, the emissions savings achieved in this application more than outweigh those costs. 

This type of hands-on collaboration with industry partners is central to how OFFIS operates. For instance, another of Naila’s projects involved assessing how manufacturers can drive down the emissions of trailers in trucking. The use of resources such as electricity and water, as well as the plant’s productivity, was measured and displayed on a dashboard at the site, giving manufacturers instant insight into the current situation on the shopfloor. The carbon footprint of the end product, as well as the company as a whole, can also be calculated from its resource use and displayed on a dashboard.

“At the end of a project you can see what materialised from it, and our work is normally implemented immediately, sometimes even before the project ends,” says Naila. By bridging the gap between academia and industry, the applied research at OFFIS also provides a path for researchers who might want to someday move to the private sector. It also means that SMEs, who may not have the expertise and resources themselves to invest in new and cost-intensive infrastructure, can benefit from the institute’s innovations to drive down their emissions. 

As well as its close relationships with both local and international industry partners, OFFIS is also distinct in that researchers are divided into Competence Clusters: groups of researchers from interdisciplinary backgrounds who collaborate around a common theme. “We can write grants together, and do research across groups as well as departments,” says Naila. “We pool our knowledge and gain insights from a variety of perspectives - ranging from experts in manufacturing systems and energy and health informatics to researchers who study the broader societal impacts of computer science.”

This approach creates plenty of opportunities for researchers to share knowledge and help develop each other’s skills. Naila describes how a colleague, an expert in user experience, helped her understand how to build a better, more user-friendly dashboard. “We also exchange ideas about methods - there may be better methods out there for what we are developing that we hadn't thought of,” she says. 

More structured forms of training are also on offer to help researchers improve their skills in areas like time management, writing, and presenting at conferences. “It has really helped me write better papers and be more confident speaking in front of audiences,” says Naila. 

The free flow of knowledge at OFFIS is underpinned by the institute’s flat hierarchy. “It's really easy to talk to my supervisor and share ideas, and there is a lot of freedom in deciding what we want to do,” says Naila. This collegiality is also evident in the institute’s support of female researchers, who are still underrepresented in the field of computer science. “There are a lot of female scientists at the institute who know how difficult it can be, and understand the kinds of challenges we might face. We have our own network and occasionally organise events for women in science, as well as for young aspiring scientists."

Beyond the walls of the institute itself, the area of Oldenburg also “has a tight-knit community,” says Naila. “Because it is such a student city, it is a really good environment for someone who wants to do research.”