Simulation-based Safety Testing of an Automated Driving System (ADS)
Name
Laima Anna Dalbina
Abstract
An Automated Driving System (ADS) is challenging to test due to the inherent complexity of the interaction between its components. Moreover, the real-world traffic in which an ADS is expected to perform is difficult due to the variety and highly dynamic interaction with other traffic agents. These systems have to be tested to ensure that the ADS maintains a desired level of safety during real-time operations. Testing an ADS in the real world incurs high costs and might be risky due to dangerous and unpredictable situations. This thesis focuses on an approach that uses simulation to test an ADS. The contributions of this thesis are two-fold. First, a method for simulation-based safety testing of ADS using the CARLA simulation environment is developed. The second contribution is the design and execution of a set of experiments to validate the method, whose results have been analyzed. The experiments were used to evaluate the ADS autonomy software currently under development in the Autonomous Driving Lab (ADL) of the University of Tartu. The thesis discusses the results of the experiments and the limitations of the developed method for simulation-based safety testing.
Graduation Thesis language
English
Graduation Thesis type
Master - Software Engineering
Supervisor(s)
Dietmar Alfred Paul Kurt Pfahl
Defence year
2024