Scenario Generation and Simulation for Autonomous Software Validation

Simulation and testing of Advanced Driving Assistance Systems (ADAS) and Autonomous Driving Systems (ADS) has gained in the recent years a lot of attention both from the automotive industry and academia. Testing this systems on the road is closer to their final real-world application and it is a desirable approach, but it is incredibly costly at the same time. Also, it is unfeasible or too dangerous to cover rare corner cases using such real-world testing. Thus, the common goal of the industry is to evaluate the systems' performance in some well-designed challenging scenarios, a.k.a. scenario-based testing, in a safe virtual environment. Testing in a virtual environment offers several advantages, including cost and time savings, as well as the ability to perform comprehensive tests that may not be feasible in the real world. Furthermore, it allows us to address the rare corner cases that are too challenging to cover on roads or proving grounds. Simultaneously, virtual testing environment allows the data generation and collection from sensors within the simulation and the testing of algorithms. This approach provides a popular alternative for evaluating the performance of ADS and ADAS in well-designed challenging scenarios, also known as scenario-based testing. This thesis primarily focuses on the design of the scenarios that cannot be effectively covered in a real-world environment. It begins by introducing some fundamental concepts related to ADAS and vehicle automation. Additionally, it discusses the regulations governing the testing of those systems in different countries. Following these regulatory guidelines, a set of relevant scenarios that meet provisions of relevant laws is proposed. An important step of the activity, involved the selection of a suitable simulation software for testing. During this process, various simulation software products available in the market are analyzed, considering their advantages and disadvantages. A tool that aligns with the requirements to conduct the testing is selected and the design and validation of the scenarios for testing is performed. The final objective is to create a set of relevant scenarios and execute them in the virtual environment. As a final step, the tool of scenario engineering is introduced to enhance the diversity and safety of real-road testing by integrating virtual and physical components in the testing loop.