Robust perception in highly automated driving under varying environmental influences
Bachelor’s/Master’s Thesis, Student Job, Forschungsmodul
We are currently looking for motivated students interested in contributing to several research projects in the area of highly automated vehicles including: robustness optimization and evaluation of perception algorithms, environment simulation and cooperative perception.
Highly automated or autonomous vehicles must be able to perceive their surroundings correctly even under difficult weather conditions. Only with a correct and complete knowledge of its environment an autonomous vehicle is able to plan safe manoeuvres without harming its environment. Therefore, algorithms and perceptual systems for local and cooperative perception must be investigated and optimized with respect to their robustness against varying environmental conditions.
Environment-aware cooperative perception pipeline
Robustness optimization of CNN-based object detection
The following list gives an incomplete overview of projects which can be worked on:
- Investigate and improve perception algorithms for resilence against environmental conditions
- Improve realism of raindrop simulation (remove drops by windshield wiper, create drops over time, …)
- Realism validation of simulated weather effects such as snow or fog
- Fusion of perceived drivable corridors of cooperative automobiles
- Implementation and evaluation of association, matching and fusion algorithms for cooperative perception
If you are interested in one of the following topics, this job may well be what you are looking for.
- Automated vehicles
- Virtual environments
- Computer vision
- Modelling & investigation of environmental influences on perception
- Object tracking & fusion algorithms
- Sensor modelling
You should have knowledge in C++ or be open to acquire basic skills in C++. Most importantly, you should be interesed in one of the listed topics!