Advancing Automated Driving Safety - Improving Particle Simulation for Risk Calculation

Institut

Lehrstuhl für Fahrzeugtechnik

Typ

Bachelorarbeit /

Inhalt

experimentell / theoretisch /  

Beschreibung

Advancing Automated Driving Safety - Improving Particle Simulation for Risk Calculation

Safety is a key aspect for automated driving. It significantly increases development resources and costs, but remains crucial. At the chair of automated driving, we are using our research vehicle EDGAR and its digital twin vEDGAR to develop and test new automation algorithms, including safety algorithms. While there is an existing safety layer, the research into online safety algorithms (such that the car evaluates the safety during driving) is still ongoing. One key method of this approach is the self-awareness, creating an understanding of both internal and external factors on the system safety. You will contribute with your thesis in this area.

Problem Formulation

Currently, a particle approach is used to evaluate the potential crash risk in a current scene. The particles represent potential traffic participants that might crash with the ego vehicle. Based on certain assumptions, the reaction of the ego vehicle to these other particles is calculated. In your work, you will investigate the current limitations of this approach and improve these. At the end, the improved particle description will be compared against datasets, a test track and the previous implementation.

Possible Approach:

You will investigate the performance of the approach along a certain test track with challenging scenarios. These scenarios can be derived from known limitations of this approach. Based on your findings, you will concretize these limitations and research approaches to improve the particle simulation for this specific observation. Ultimately, you will test your algorithm on the test track again and evaluate it against the previous algorithm. In addition, the behavior of your particles will be compared to the behavior of actual vehicles in datasets.

Voraussetzungen

You need interest into the topic of simulation and automated driving and experience with Python or C++. I am looking for students who have fun finding their own solutions for a problem and developing their own approaches. Experience with CARLA is of help, but not a must. If you are interested, please reach out to me via: nils.gehrketum.de with your transcript of records.

 

 

Verwendete Technologien

CARLA, Simulation, Automated Driving, C++, Python

Tags

FTM Studienarbeit, FTM AV, FTM AV Safe Operation, FTM Gehrke, FTM Informatik

Möglicher Beginn

sofort

Kontakt

Nils Gehrke, M.Sc.
Raum: MW 3507
Tel.: +49 89 289 10342
nils.gehrketum.de

Ausschreibung