Affiliated Research Projects

Maßnahmen zur Transparenzerhöhung der Bewegungsabsicht von Robotern in Abhängigkeit menschlicher Aufmerksamkeit (MaTraB)

ESF-Plus-Nachwuchsforschungsgruppe 

Ansprechpartner: 

Prof. Dr. Dr. Marco Ragni | Dr. Kristina Kittner
Professur: Prädiktive Verhaltensanalyse
Fördermittelgeber: Sächsische Aufbaubank (SAB)
Laufzeit: 01.11.2025 - 31.10.2028

Das Projekt untersucht, wie Roboter ihre Bewegungsabsichten für den Menschen transparent, verständlich und situationsgerecht kommunizieren können. Im Fokus stehen unterschiedliche Kommunikationsmechanismen sowie deren Kombinationen und deren Wirkung auf Transparenz, Vertrauen, Effizienz und Robustheit der Mensch-Roboter-Interaktion in Abhängigkeit der menschlichen Aufmerksamkeit.Hierzu werden neuartige, aufmerksamkeitsadaptive und energieeffiziente Algorithmen entwickelt, mit denen Roboter ihre Bewegungsabsichten beispielsweise visuell oder sprachlich vermitteln können. Die Ansätze werden in aufeinander aufbauenden Experimenten evaluiert und so gestaltet, dass sie auf verschiedene Robotersysteme und reale Anwendungsszenarien übertragbar sind.Das Projekt adressiert zentrale Herausforderungen insbesondere für produzierende KMU, indem es flexible, nutzerzentrierte und nachhaltige Robotiklösungen ermöglicht. Darüber hinaus fördert es den Wissens- und Technologietransfer, stärkt den Forschungsstandort Chemnitz und unterstützt die Qualifizierung von Nachwuchswissenschaftlerinnen im MINT-Bereich.

Projektbeteiligte: 

Prof. Dr. Florian Röhbein | Professur Neurorobotik
Prof. Dr. Angelika Bullinger-Hoffmann | Professur Arbeitswissenschaft und Innovationsmanagement
Prof. Dr. Fred Hamker  | Professur Künstliche Intelligenz
Prof. Dr. Harlad Kuhn | Professur Smart Systems Integration
Prof. Dr. Wolfgang Einhäuser-Treyer | Professur Physik kognitiver Prozesse

Kognitive Modellierung

Die Fähigkeit zur Schlussfolgerung aus Informationen ist ein Hauptforschungsbereich der Kognitionsforschung. Traditionelle Untersuchungen fokussieren auf räumliches, konditionales und syllogistisches Schlussfolgern. Komputationale Modelle gewinnen an Bedeutung, um menschliches Verhalten präzise zu erfassen. Ein Ziel ist die Entwicklung von Prozessmodellen, um Denkprozesse zu formalisieren und experimentelle Daten zu vergleichen. Bisher fehlen jedoch Modelle, die über Domänengrenzen hinweg und individuelle Unterschiede berücksichtigen und präzise Vorhersagen liefern können. Dieses Projekt zielt darauf ab, existierendes Wissen aus Theorien in ein einheitliches kognitives Modell zu integrieren. Dabei sollen interindividuelle Unterschiede berücksichtigt und Prozessmodelle automatisch generiert werden, um hohe Vorhersageleistung zu erreichen und sich an individuelle Unterschiede anzupassen. Die Methode ist frei von zusätzlichen theoretischen Annahmen und zielt ausschließlich darauf ab, die Vorhersagekraft zu verbessern, während die Erklärbarkeit durch den Ursprung der kognitiven Operatoren in Theorien gewährleistet bleibt. Das entwickelte Framework soll nicht nur das Schlussfolgern, sondern allgemein die Modellierung menschlicher kognitiver Prozesse vorantreiben.

Ansprechpartner:

Prof. Dr. Dr. Marco Ragni
Professur: Prädiktive Verhaltensanalyse
Fördermittelgeber: Deutsche Forschungsgemeinschaft
Laufzeit: 01.01.2024 - 31.12.2025

MINT-Fit – Inspiring STEM in Southwestern Saxony

The MINT-Fit project targets primary school children growing up under challenging circumstances—such as in educationally disadvantaged families, with refugee experience, or in structurally weak regions. Its aim is to provide continuous, age-appropriate access to STEM subjects (Mathematics, Computer Science, Natural Sciences, and Technology), opening up new educational opportunities and supporting their holistic personal development.

The project connects institutions that either already work with children from these target groups or possess recognized STEM expertise. Together, they develop low-threshold, hands-on programs designed to spark curiosity, strengthen skills, and foster long-term enthusiasm for scientific questions.

MINT-Fit is led by Chemnitz University of Technology in close cooperation with regional partners: Don Bosco Sachsen, KINDERVEREINIGUNG Chemnitz e. V., the Johanneum Children and Youth Foundation of the City of Chemnitz, and Phänomenia Stollberg. The project is funded over three years with more than €500,000 by the Federal Ministry of Education and Research.

Planned activities include STEM clubs, holiday camps, experiment days, and science shows, all pedagogically tailored to the needs of primary school children. A special highlight is the MINT-Mobile—a mobile learning space that travels throughout southwestern Saxony as a rolling discovery workshop, reaching children in remote locations and inspiring their interest in science and technology.

Contact: Prof. Dr. Leena Bröll
Chair: Primary Education – Science Teaching
Funding: Federal Ministry of Education and Research
Duration: Until June 2027

Systems, Algorithms, and Cognitive Models for Predicting Individual Human Reasoning Processes (PREDIR)

This research project addresses a central question in cognitive artificial intelligence: How can we predict the conclusions an individual will draw based on only a few observed pieces of information? Unlike classical approaches in cognitive psychology, which primarily focus on average behavior, this project specifically targets individual thinking. The goal is to develop computational models that adapt to individual persons and can predict their future cognitive decisions—even with limited data.

The project integrates perspectives from cognitive science, psychology, and data-driven methods such as recommender systems. It analyzes existing approaches, identifies their limitations, and develops improved models based on these insights. The focus is on both the performance of predictive systems and a deeper understanding of the cognitive mechanisms underlying human thought.

The long-term aim is to create adaptive, learning-capable AI systems that can not only interpret human thinking but also anticipate it on an individual level.

Contact: Prof. Dr. Marco Ragni
Chair: Predictive Analytics
Funding: German Research Foundation (DFG)
Duration: 01 July 2024 – 30 June 2027 (36 months)

Automatic Process Model Generation for Cognitive Modeling

This project aims at a paradigm shift in cognitive modeling: Instead of using entire theories to describe group behavior, the smallest building blocks of existing theories—so-called atomic cognitive processes—are identified, combined, and individually adapted. Using a formalized space of cognitive operations and accounting for individual differences, the project seeks to automatically generate cognitive models that are highly accurate, individually customizable, and powerful in prediction. This innovative methodology prioritizes predictive capability over theoretical assumptions, opening new avenues for cognitive research on human reasoning.

Contact: Prof. Dr. Marco Ragni
Chair: Predictive Analytics
Funding: German Research Foundation (DFG)
Duration: 01 January 2024 – 31 December 2025 (24 months)

HYDRA – Hybrid Decision Support for Computer-Aided Design

Mechanical design places high demands on engineers: in addition to mechanical stability and safety, aspects such as aesthetics, environmental impact, maintenance effort, costs, and recyclability must also be considered. Many of these criteria are difficult to formalize, which is why human intuition and experience continue to play a central role.

The goal of the HYDRA project is to enhance the design process through intelligent, hybrid decision support. Classical CAD systems are combined with an AI-based assistance system, the so-called Completer. Trained using real-world design data, the Completer proposes step-by-step new design options based on the current geometry and individual inputs from the engineer.

Through an interactive overlay, targeted feedback can be provided and specific constraints set—for example, to keep certain parts of the design unchanged or to suggest modifications. Humans remain at the center of the process, retaining full control and decision-making authority, while being effectively relieved of routine tasks.

This type of support is particularly valuable for less experienced designers, helping them achieve high-quality design solutions more quickly and efficiently.

Contacts:
Prof. Dr. Fred Hamker – Chair of Artificial Intelligence
Prof. Dr. Alexander Hasse – Chair of Machine Elements and Product Development

Funding: German Research Foundation (DFG, SPP 2443)
Duration: 01 January 2025 – 31 December 2027 (36 months)