The project delivered a portfolio of demonstrators that showcase the practical power of artificial intelligence across industrial, educational, and research settings. In the collaborative‑robot demonstrator, developed jointly with the Handwerkskammer Dresden, a worker’s identity is verified through facial matching, while real‑time emotion and fatigue detection feed back to the robot’s control system. The system relies on a multi‑GPU cluster for training, an Nvidia Jetson for on‑board inference, and high‑resolution cameras for visual capture. The demonstrator was installed in both the IIoT testbed and the HWK Dresden workshop, allowing live demonstrations of AI‑enabled safety and productivity enhancements.
The chess‑robot upgrade illustrates the benefits of synthetic training data. By generating a diverse dataset entirely within Nvidia Omniverse, the team eliminated the need for labor‑intensive image collection. The resulting convolutional neural network achieved high precision in piece recognition even under variable lighting, outperforming the earlier classical image‑processing baseline. The same hardware stack—multi‑GPU training, Jetson inference, and a camera—was used, and the demonstrator was showcased at public events to highlight the potential of synthetic data for robust vision systems.
A Lego‑sorting machine demonstrates AI in a time‑critical production scenario. Falling Lego bricks are captured by a high‑speed camera and classified by a neural network running on a Jetson, after which an actuator directs each piece into the correct bin. The demonstrator emphasizes that modern inference hardware can deliver decisions within strict latency budgets, making AI viable for real‑time manufacturing tasks.
In automotive quality control, a miniature assembly line model was built to inspect vehicle components. The system performs part detection and defect identification using a model trained exclusively on synthetic data, again reducing data‑collection effort while maintaining high detection accuracy. The demonstrator also serves as a platform for future extensions such as predictive quality monitoring, although some planned features were delayed due to hardware procurement issues and the COVID‑19 pandemic.
Additional planned demonstrators—an interactive museum tour guide robot, a robust service‑robot platform for patient care, and a land‑equipment monitoring system—were not fully realized within the project period. Delays in hardware delivery and restricted access to partner facilities limited the scope of these prototypes, but the specifications were documented for future student projects and research work.
The project’s technical outcomes are complemented by a strong educational impact. The HTW Dresden leveraged the newly acquired GPU resources to enable all students to experiment with complex AI models during lectures and theses. Several master’s and bachelor’s theses were supervised, covering topics from thermography for dairy herd health to deep‑learning analysis of smart‑meter data, and from model transfer for fault detection to infrastructure‑as‑code for Kafka clusters. These works benefited from the multi‑GPU cluster, Jetsons, and peripheral equipment, as well as from the expertise of industry partners ranging from startups to large corporations.
Collaboration was central to the project’s success. The primary partner was the HTW Dresden, which coordinated the overall effort and managed the hardware infrastructure. The Handwerkskammer Dresden contributed domain expertise and workshop space for the collaborative‑robot demonstrator. The TU Dresden ZIH supplied the multi‑GPU cluster and facilitated integration with the university’s research ecosystem. OncoRay and other research institutes participated in the development of the data‑science tool for tumor pathology. Funding came from multiple sources, including the Federal Ministry of Education and Research (BMBF) under project codes 03ZK212AC and 16MEE0262, the Federal Ministry of Economic Affairs and Climate Action (BMWK) under FKZ 13IK021E, and the State Ministry for Science, Research, and the Arts (SMWA/SAB) through several calls such as ZW9NDGT2Z and MMEGBWHWY. The project spanned from 2021 to 2025, during which the HTW Dresden expanded its research capabilities, secured further external funding, and established a foundation for interdisciplinary AI research and industrial collaboration.