Scene understanding and reliable deep learning for robotics

Study level

Vacation research experience scheme

Topic status

We're looking for students to study this topic.


Dr Niko Suenderhauf
Senior Lecturer
Division / Faculty
Science and Engineering Faculty


For many practical applications of robotics and embodied AI, such as driverless cars, healthcare and domestic service robotics, scene understanding is a critical component. Modern methods rely heavily on deep learning to extract information about the scene from camera images.

Research activities

You will support our ongoing research in the areas of:

  • semantic scene understanding and semantic SLAM
  • probabilistic object detection
  • reliability of deep learning for robotics applications
  • reinforcement learning based on semantic information.
Based on your interests, we can tailor the project to be in a certain area of research. Your role may be software-heavy or involve working with real robots.


Depending on the chosen project topic potential outcomes may include a software module that demonstrates a certain capability on a robot or a dataset, or a component that supports the research of one of our PhD students and can be used in a publication (evaluation, experiments, etc.).

Skills and experience

Depending on your chosen topic the requires skills vary.

In general you should be good at programming in Python and have a good understanding of linear algebra.


You may be able to apply for a research scholarship in our annual scholarship round.

Annual scholarship round



Contact the supervisor for more information.