SA/BA/MA: Data Analysis & Machine Learning with Hidden Markov Models in Driving Behavior (Automated Driving)

08.07.2026, Abschlussarbeiten, Bachelor- und Masterarbeiten

Background:
Within the research project MiRoVA, we investigate interaction behaviors in mixed-traffic environments, specifically focusing on scenarios where automated vehicles (AVs) and manually driven vehicles coexist. Utilizing data collected from advanced driving simulators and eye-tracking systems, this thesis aims to analyze action sequences and interaction patterns between three key actors (see Figure). The core objective is to model these dynamics using advanced statistical and mathematical approaches to understand how human drivers adapt to autonomous systems compared to other human drivers.

Responsibilities:
• Sequence Modeling: Model and analyze sequential action data using Hidden Markov Models (HMMs) or similar methods.
• Interaction Modeling: Apply Game Theory frameworks to model and quantify driver-AV and driver-driver interactions.
• Comparative Analysis: Compare interaction dynamics between driver-AV dyads and driver-driver dyads to identify key behavioral shifts.
• Pipeline Development: Build a structured, well-documented, and reusable data analysis pipeline.
• Visualization & Documentation: Visualize complex behavioral data and document your findings in a high-quality scientific thesis.


Requirements:
• Degree program in Human Factors Engineering, Mechanical Engineering, Computer Science, Mathematics, Data Science, or a related field
• Strong foundation in data analysis, statistics, and probabilistic modeling
• Solid understanding of Machine Learning, with specific experience (or strong interest) in Hidden Markov Models or other Markov processes.
• Advantage: experience with time series, sensor data, or driving data

Kontakt: tianyu.tang@tum.de

Mehr Information