Analysis of Advanced Driver Assistance Systems (ADAS) to improve traffic safety of autonomous vehicles

The present master thesis has the purpose to analyze Advanced Driver Assistance Systems (ADAS) regarding their potential to improve traffic safety of autonomous vehicles. First the characteristics of the automated driving levels are described and ADAS as active safety systems are differentiated from other systems. Several ADAS are defined based on their functionality and are categorized based on the automation levels. A New Era in safety technology is exceeded by the introduction of partially automated driving concepts since 2016. Although the number of fatalities is decreasing, the number of accidents is increasing year-after-year, whereby nearly 90% are caused by human mistakes. Studies for the theoretical potential of collision avoidance and the consumer acceptance revealed, that the Autonomous Emergency Braking (AEB) systems is superior to the other ADAS. By conducting a market analysis on the new German car market about the availability of ADAS, it was found, that AEB systems are superior too. AEB systems are offered for 69 % of all models. In general, two versions are distinguished: AEB city for small speed ranges, and AEB interurban up to the top speed, whereby both can be offered in combinations to the customer. ACC LKA and PA are available for around the half of the models. An in-depth analysis of the AEB examined technical details on how the risk assessment of AEB systems is done by a patent search and with published information of the OEM. Thereby, it is found that most of the OEM seems to work with risk assessment metrics TTC and ETTC and today's AEB is designed to prevent human failure. But in case of autonomous systems, system failure must be prevented from the system itself and therefore, the system architecture has to be changed in the design. In the state of the art the most important closed and still ongoing projects in the field of autonomous driving and safety, as well as strategic partnerships of companies for collaborating research and finally some concept cars of well-established OEM and the vehicles of new entrants are described. Main fields of research affect the concepts to combine the human driver and the autonomous driving system and the application of AI. Finally, the Safety Supervisor (SSV) concept, a framework for runtime safety monitoring of autonomous systems, is explained and exemplified on a platooning system in Matlab/Simulink. This is done by implementing an AEB system with three risk assessment metrics TTC, ETTC and TTB. Results of simulations have shown, that the AEB is able to avoid, if TTC or ETTC are used, or mitigate, if TTB is used and therefore, the SSV can provide a useful framework for future research