An important goal of the integrative research at the CoR-Lab is to further develop functions and development methods for intelligent socio-technical systems in workplaces of the future. Exemplary challenges are to easily and safely adapt the behavior and support functions of cognitive systems to changing tasks, different environments and people with their individual characteristics. Model-driven software and system development, the application of data-driven learning methods from AI research and their integration in hybrid approaches are important building blocks for this. In the following, some examples of current research projects are shown whose methods and technologies are representative for application in practice.
In the Competence Center Arbeitswelt.Plus, CoR-Lab is developing an AI-based core function for the smart laundry together with Herbert Kannegiesser GmbH. The objective of the project is to minimize the contact of the working human with soiled and possibly contaminated laundry. This is achieved by automatic classification and sorting of the delivered laundry with machine learning methods, so that the classified laundry can be treated as clean as possible without direct human contact. Current approaches of "eXplainable Artificial Intelligence" (XAI) are adapted and further developed to the needs of employees in medium-sized companies, so that a new division of labor between humans and AI-controlled machines is created.
The VEDLIoT project, funded by the European Union and coordinated by CoR-Lab, is developing a novel IoT platform that uses deep learning algorithms distributed across the IoT continuum. The project offers a low-threshold funding program at project half-time to engage additional companies by evaluating the IoT platform against additional industrial use cases in focused projects. Through the application of the IoT platform, the market readiness of the VEDLIoT solutions will be demonstrated and further increased through feedback from the additional companies involved.
The joint project ITS.ML, which is funded by the BMBF, focuses on the use of machine learning methods for intelligent technical systems of the next generation. Machine learning (ML) methods and tools are to be developed, implemented and tested to enable the use of current ML technology, especially in regional small and medium-sized enterprises. A practical example from the project is a methodology for using prior knowledge in combination with reinforcement learning for the automated assembly of complex components. The use of model-based prior knowledge reduces the effort required for the application of learning methods and supports efficient adaptation to new assembly or disassembly tasks.