Our research focuses on algorithms and systems for processing and interpretation of sensor data. A special emphasis is on the combination of machine learning with models and prior knowledge. To achieve this, we use combinations of engineering (i.e. probabilistic methods and algorithm design), machine learning, and symbolic AI approaches (e.g. Vector Symbolic Architectures / Hyperdimensional Computing).

Application areas include mobile robotics, automotive, automation, and wherever sensor data processing and interpretation is valuable including industrial applications, human-machine interaction, health care, rehabilitation, and medical applications.

What kinds of problems are we trying to solve?

• reliable and robust visual perception in challenging and changing environments
• 4-D reconstruction of dynamic scenes providing 3-D models over time
• AI-based systems that offer a high degree of control, security, and understandability
• recognition, classification, and detection
• (sensor) data fusion, analysis, and interpretation
• incorporation of model and prior knowledge in learning-based systems
• time series analysis
• decision-making and action execution

For a list of our publications please see here.