The 20th and 21st joint COTA International Conference of Transportation Professionals (CICTP 2020-21)

With the continuous development of autonomous driving, the test methods of traditional automobile can’t satisfy the validation of autonomous vehicle. Different from the traditional vehicles, a scenario-based test method is adopted in the validation of autonomous Driving. For each test scenario, it is a challenge to generate test cases that cover the complex and varied real-life traffic. Vehicle cut-in scenario is a common but risky scenario of real traffic on highways. In this paper, multiple real samples of vehicle cut-in scenario were extracted from highD dataset that is a high-precision vehicle trajectory database on highways. By analyzing of motion parameters and the positional relation between participated vehicles from these real samples, a description model of vehicle cut-in scenario was built to generate test cases. In addition, the risk degree of this scenario was evaluated depending on TTC in cut-in point. At last, combining with the distributions of parameters in description model, test cases generation was executed using Monte Carlo method. Simulation results of test case generation shows that the generated cut-in test cases are capable to cover all risk level, which provides a large number of test cases to support autonomous driving testing. Keywords: Testing of autonomous vehicle; vehicle cut-in scenario, Test cases generation; Monte Carlo method

CICTP