Edge AI Research Group

The EdgeAI working group focuses on systems at the edge of the network that do not require a permanent connection to the cloud or a server. The aim is to implement AI applications directly on these platforms, including embedded AI. This involves resource-constrained target systems – often small microcontrollers – on which powerful AI-based applications are implemented using suitable methods.

Copyright: THU

Research topics

  • Generation of artificial time series data for energy consumption
  • Prediction methods for data-poor systems based on time series
  • Light RAG for low-resource target systems
  • Computer vision for smart glasses

Projects

MIKE: More inclusion through smart AI on end devices
Copyright: THU

AI and augmented reality are used to promote inclusion and support for people with disabilities. The central unit is smart glasses, for which the working group is implementing special support functions such as barcode recognition.

Further information on the MIKE project

FRIDA: Robot dog as an assistant for people with disabilities
Copyright: THU

The robot dog serves as an assistant for people with disabilities and in specific hazardous situations. It was developed using AI-based image processing and sensor data processing.

More information about the FRIDA project

BEMS: Balcony power plant management system
Copyright: THU

The goal of the project is to achieve the highest possible level of self-sufficiency for households with plug-in solar devices, using a small amount of user data and appropriate machine learning algorithms.

More information about the BEMS project

Assistant KAI
Copyright: THU

The humanoid robot KAI is being developed into an assistant for all questions related to the university.

More information about the KAI project

Topics for theses

The topics for final theses are published on the intranet of the Faculty of Electrical Engineering and Information Technology. There, students will find an up-to-date overview of available topics as well as further information on the respective content.

If students are interested in a topic, they can get in touch with the relevant contact person directly.

Contact