Eye Tracking and AI-Based Support Systems
Eye-tracking records where a person looks, how long their gaze rests on a given area, and in what order it moves across a screen. It is a powerful tool for gaining insight into underlying cognitive processes, as gaze patterns can reflect how people attend to information, make decisions, and learn. Eye-tracking is used across many fields; our focus lies in education, and in mathematics education in particular, where students who arrive at the same answer may have taken very different solution paths. Equally important is the analysis: eye-tracking produces large and multimodal data, and we use AI to analyze these gaze patterns automatically and to derive decisions from them, such as identifying a student's strategy and selecting suitable follow-up support.
KI-ALF
Adaptive learning system for mathematics education, powered by eye tracking and AI
Eye movement data captured with two different eye tracking devices
Eye movement data captured with dedicated eye tracking device
Fixation map representation of eye tracking data
Heatmap representation of eye tracking data
Projects
Mehr Informationen
One of our projects, KI-ALF, is an adaptive learning system designed to support mathematics education. KI-ALF combines eye-tracking with artificial intelligence, using eye-tracking data from students as they work on mathematical tasks covering basic mathematical competencies. Low-cost eye-tracking is realized through a consumer-grade webcam and AI, and the system analyzes these gaze patterns to diagnose and support individual mathematical basic competencies, aimed especially at students at the transition from primary to secondary school who need additional support. Teachers can follow eye movements and task processing live on a second screen and step in with additional help, while the system records the competence analyses and outputs the results in a report. KI-ALF is explicitly developed for practical deployment in schools and everyday classroom settings, and the project is funded by the German Federal Ministry of Education and Research (BMBF).
Another project we worked on was MADITA. The MADITA project develops, evaluates, and disseminates a digital app to support students in Grade 1 in early math skills. It addresses the need to support students’ skills early on in schooling. The app will be research-informed and use AI and eye-tracking to identify each child’s individual competence profile and their risk to develop math difficulties and to provide customized support for each child at their respective level. It will allow for teachers to monitor and evaluate students’ learning process with the app. MADITA facilitates digitalized, individualized, adaptive support of early math skills. The MADITA app will diagnose early math abilities, generate customized support sequences, and provide meaningful, directly usable individual student profiles to teachers.
Affordable eye-tracking for schools
Eye-tracking has long depended on specialised, expensive hardware, which has largely confined it to laboratory research. A central interest of our work is making the method affordable enough to bring into schools, using ordinary webcams instead of dedicated devices. This requires two things: reliable gaze prediction from low-cost cameras, and real-time eye-tracking analysis of the resulting data, so that diagnostic information and adaptive support are available while a student is working rather than only afterwards.
Our research interests lie at the intersection of eye-tracking, artificial intelligence, and education. In particular, we work on:
Gaze prediction from standard webcams as a low-cost alternative to dedicated eye-trackers
Eye-tracking analysis using AI, including real-time analysis during task processing
Foundation models for eye-tracking analysis, learning general patterns across large amounts of gaze data
AI-based adaptive learning support, using gaze-based diagnostics to select suitable support for individual students
Applications in mathematics education, particularly for students who need additional support at the transition from primary to secondary school
In case you are interested in our work, please do not hesitate to contact me, parviz.asghari(at)tum.de
Our Team
Dr. Han Fan, Postdoc han.fan(at)tum.de
Parviz Asghari, PhD candidate working on an intelligent and affordable eye tracking-based support system for mathematics education parviz.asghari(at)tum.de