Seed project:Automated Assembly/Disassembly Instructions (ADDITION)

Project information

Project manager
Welf Löwe
Other project members
Osama Ziada, Diana Unander, Linnaeus University; Marie Jonsson, Mehdi Tarkian, Linköping University; Jason Brim, Sigma Technology; Wouter Maagdenberg, TXTOmedia
Participating organizations
Linnaeus University, Linköping University, Sigma Technology, TXTOmedia, Volvo CE
Financier
Linnaeus University Centre: DISA
Timetable
1 June 2023 – 30 November 2023
Final report
Final report ADDITON.pdf
Subject
Computer Science (Department of Computer Science and Media Technology, Faculty of Technology)

More about the project

Work instructions created and used for the 'factory floor' can often be leveraged for maintenance, but nowadays often rest in separate content silos in the company (production vs support). Reuse & share is a valid option to stimulate coherence and compliance. Work instructions and information describing how to perform assembly, disassembly or maintenance are today manually created by using written documentation (like standards, procedures etc.), CAD models and engineering knowledge. With sustainability, industry must adapt to the circular economy, meaning more products must be maintained, assembled, and disassembled to perform restorative actions like remanufacturing or supporting re-use of components. With the current manual approach, this is time-consuming and costly, which will impact the competitiveness of companies that want to implement circularity as part of their product offering.

Currently, manufacturers create service and maintenance information based on product design, experience, CAD files, and other input from suppliers. This might mean that a simple maintenance instruction is created based on the experience of the content creator with help from CAD files and verification against additional input for things such as tolerances. Initial training of the factory, support content for the factory on the one hand, and training and support content for the aftermarket could be leveraged, including the instructions for end users for certain tasks. It is a time-consuming, manual process, both in creating relevant text descriptions and exporting usable images.

In the automotive industry, creating content for a new vehicle some text might be reused from previous products, but new, descriptive images are needed. How much can be reused, however, is dependent on the software and tools used, and in many cases, everything has to be recreated. When it comes to end-of-life/disassembly instructions, the current understanding of the consortia is that these are non-existent, because the manufacturer’s focus is on selling and maintaining vehicles rather than the disassembly/end-of-life. Also, the role of dealerships changes rapidly. Cars are more and more sold directly to the market instead of via the dealership. The income of a dealership relies more and more on services, and options are open to more (non-branded) dealership, who need to be able to access information on how to service and dissemble the parts/car. Information that used to be shared within the own channel, has/should now become public. Some information might be shared with end-of-life/disassembly partners concerning which parts can be reused and resold, but otherwise not much is traditionally created, and nothing is shared publicly. With the current manual approach, the creation of work instructions is a time-consuming task and thus drives costs, making companies that want to take the circular approach less competitive. Also, demands from the EU and Swedish government on circularity will increase the need for more efficient methods to create work instruction to support the transition to a more sustainable society.

With rapidly developing AI methods available, these can be utilized to automatically create needed information, supporting companies in offering circularity as part of their solution portfolio, but remaining cost competitive. Also, this supports the recirculation of resources, and thus environmental sustainability according to Agenda 2030. To get started with the research and the consortium we apply for this seed project; its objectives are to:

• Map the state-of-the-art for the field and identified gaps to fill
• Gather information on and examples of the data available from our potential partners.
• Train Artificial Intelligence (AI) models—based on the state of the art and the data available—to generate prototype work instructions for assembly, disassembly, or maintenance. This constitutes the main part of the work in the seed project. We will create different types of instructions: texts, image sequences, videos, and mixtures thereof. It will serve as a feasibility study to the ambition level for the proposal according to the state of the art.
• Link interactive instructions with (spare) part (number)s.
• Componentize content to stimulate reuse. Parts are often used in various products, or are an integrated part of a bigger procedure.
• Create a consortium for a project within automation of work instruction creation, with the mapping of stakeholders and a deeper understanding of the value proposition of the targeted technologies.
• Develop a framework proposal for a suitable call ready to submit this autumn.

The project is part of:

• The research group Data Intensive Software Technologies and Applications (DISTA)
• The research group Smart Industry Group (SIG)

DISA's seed projects

Learn more about the seed project concept and DISA's other seed projects.

Staff