Research Interests

I am interested in developing resilient autonomy for robots operating in uncertain, unstructured, and data-limited environments. My research focuses on combining machine learning, control theory, and information-driven decision-making to enable robots to learn, adapt, and act reliably in challenging domains such as rough terrain and underwater environments.

• Learning-Based Control – Integrating methods from machine learning and control theory to develop data-driven controllers for complex dynamical systems, with a focus on robustness, adaptation, and uncertainty-aware decision-making.
• Active Learning and Informative Planning for Dynamical Systems – Developing strategies that actively collect task-relevant data to identify system dynamics, reduce model uncertainty, and improve downstream control or navigation performance.
• Perception-Aware Planning and Control – Designing planning and control methods that explicitly account for perception quality, such as visibility, observability, object detection confidence, localization uncertainty, and information gain.

Student Projects and Thesis

I am constantly looking for motivated students who are interested in my field of research. Please contact me via e-mail if you are interested in working on a thesis (e.g., bachelor or master thesis) under my supervision, even if no open topics are currently listed.

Please include your preferred starting date as well as your CV and transcript of records in your e-mail. This helps me to select a topic matching your background.

Potential Topics

The following topics are possible starting points. The exact scope can be adjusted depending on the project type and the student's background.

• Task-Oriented Active Learning for Dynamical Systems (FP/MA)
   Comparing data-selection criteria such as uncertainty-based sampling, sensitivity-based sampling, and task-oriented value-of-information methods. The goal is to evaluate active learning not only by model prediction error, but also by downstream control or navigation performance.
• Heteroscedastic Uncertainty Modeling for Robot Dynamics (FP/MA)
   Modeling input-dependent uncertainty in robot dynamics, for example residual errors that depend on terrain slope, surface type, velocity, or slip conditions of offroad vehicles. Possible directions include uncertainty calibration, Gaussian process or neural-network-based regression, and active learning under heteroscedastic noise.
• Perception-Aware Planning and Control (BA/FP/MA)
  Planning and control methods that account for perception quality, such as visibility, observability, object detection confidence, or localization uncertainty.
• Benchmarking Off-Road Mobility and Learning-Based Control with Verti-Bench (IP/BA)
   Setting up and analyzing benchmark scenarios for off-road robot mobility. Possible directions include reproducing baseline experiments, analyzing terrain difficulty and failure modes, and identifying scenarios suitable for learning-based control or active learning evaluation.