PRO-LAB – Development of intelligent task and process control for the energy-efficient use of AI systems
In the PRO-LAB project, the expansion of the research infrastructure ( as part of the IdEaL AI Space for Energy Efficiency1 project funded by the ERDF 2021-2027 funding line) and the associated improvement in computing capacities, acquires, networks, and processes a wealth of information and data from various heterogeneous sources on campus with the help of artificial intelligence (AI). The PRO-LAB project is investigating how innovative algorithms can be used to implement resource-saving and energy-efficient task distribution (queues) for generating computationally intensive AI . This research project is investigating how tasks related to the application and use of computationally intensive AI processes can be distributed over time and across different hardware. The aim is to train models only when there is a surplus of green energy available from the campus's own sustainable sources, such as the photovoltaic systems operated on campus. Due to the heterogeneous and complex infrastructure on campus and the different forms of energy (heat, cooling, electricity, and others) and measurement systems, it is currently not possible to carry out a comprehensive analysis of energy efficiency measures for the entire campus.
The project aims to process the findings obtained by PRO-LAB in such a way that they can be made available to companies and the business community in Bremen in particular.
The approach of KI-Space for Energy Efficiency (KI-SEE) aims to derive models from networked heterogeneous data that can be used to make predictions about energy-efficient measures. KI-SEE thus implements investments in hardware, while thePRO-LAB research project buildsthese systems and investigates how machine learning models can be generated in a resource-efficient manner. Particularly in the area of regional technology and knowledge transfer, as well as digital transformation through the use of innovative artificial intelligence methods, the PRO-LAB project makes modern AI methods transparent and available to scientists and the Bremen economy. The PRO-LAB project thus significantly supports key innovation in Bremen's innovation strategy. Specifically, the PRO-LAB project is researching the resource-efficient use of AI systems through intelligent queues and task distribution and integrating prototypes into the AI SEE. By researching such energy-efficient task distribution, AI systems, for example, are only to be trained when sufficient green energy is available on campus based on predefined priorities and taking into account other computing tasks on the KIClusters and servers. At the same time, energy consumption is estimated in order to make both the training operation and the subsequent use of the AI models as efficient and resource-saving as possible.
This research project therefore aims to develop energy-efficient measures through improved R&D conditions in environmental protection and the use of innovative AI methods, opening up new perspectives in both scientific and academic education and industrial applications. The findings will be incorporated into theses, dissertations, and courses at the University of Bremen and used for economic application scenarios—for example, in aviation or in the operation of autonomous drones. This will enable a wide range of stakeholders in research projects, industrial contracts, and teaching to benefit from a sustainable deepening of core competencies and the derivation of new business models.
The PRO-LAB project and the establishment of the interdisciplinary energy research laboratory at the University of Bremen will improve the research infrastructure on the Bremen campus, creating a future-oriented and modern platform for the development of innovative AI systems . The linking and promotion of the PRO-LAB research project thus contributes significantly to the goals set by the EU Commission to accelerate investment in AI technologies in order to promote a robust economic and social upswing.
