Friday, July 08, 2022
Generation of driving scenarios near realistic edges
Topic and Goal of the Thesis
Vehicles are driving increasingly automated. But how can we prove that these vehicles are safe?
This is a central question in research on safeguarding automated vehicles. A promising approach is scenario-based testing. Particularly exciting are edge cases that challenge the system but are realistic. However, such edge cases can hardly be found in data.
In this thesis, a method for the synthetic generation of such edge cases on the basis of real data shall be developed. The goal is to determine the general boundary of real-world scenarios based on real data. Rule-based approaches as well as machine learning approaches can be considered.
- Literature research on the topics of scenario generation and parameter extrapolation
- Development of a methodology to generate realistic Edge-Cases for dynamic road users
- Implementation of the method
- Validation of the methodology based on real intersection data
- Good English or German language skills
- Reliability, commitment and enjoyment of working independently as well as methodically
- Basic knowledge in data science
- Experience with python
Note: Please attach brief resume and grade summary.