Identification and visualization of the disengagement scenarios for highly automated vehicles
- Lehrstuhl für Fahrzeugtechnik
- Bachelorarbeit Semesterarbeit
While the advancement of self-driving vehicles has reached the state where vehicles can solve complex situations, they will still face scenarios that will challenge them to leave the operational design domain and eventually to come to a still-stand - making them unable to continue the operation. Such events are usually caused by uncertainties in perception, by environmental circumstances, infrastructural changes, or scenario complexity in general. Teleoperation is one promising approach to support the vehicle to overcome such challenges. It can be used to replace the need for in-vehicle fallback drivers utilizing various remote assistance interactions that not only significantly reduce the downtime but also reduce the costs and gather new data for further system improvements.
The goal of this thesis is to identify currently most relevant disengagement scenarios through the analysis of publicly available datasets, disengagement reports, and literature. The identified scenarios will be assessed by specific metrics and compiled into a catalog, that will contain not only the scenario description but also an implementation of the most relevant ones within the SVL or CARLA simulation framework. The catalog will be built from the perspective of teleoperated interaction and it is meant to support the further research of these topics at the Chair.
The following work packages will be covered:
- Literature research and assessment metrics definition
- Data analysis (Python/Matlab)
- Scenario description, visualization, and compilation into a catalog
- Scenario simulation (SVL or CARLA)
- Evaluation and documentation of the results
- Programming experience in Python and Matlab, Git
- Knowledge in the field of autonomous driving (nice-to-have)
The thesis can be written in German or English language. In case this has caught your attention, send your CV and transcript of records to majstorovicftm.mw.tum.de.
- FTM Studienarbeit, FTM Majstorovic, FTM Automatisiertes Fahren
- Möglicher Beginn
Domagoj Majstorovic, M.Sc.
Raum: MW 3507