From operational design domain to test cases: A methodology to include harsh weather

Abstract

Background To gain widespread use, assisted and automated driving (AAD) systems will have to cope with harsh weather conditions, such as rain, fog, and snow. This affects the development and testing of perception and decision-making systems. Since the weather cannot be controlled in field tests, the availability and use of virtual simulation and test facilities that can accurately reproduce harsh weather becomes vital. Test cases subjecting the system under test to harsh conditions, covering all expected weather phenomena in both typical and challenging scenarios, must be defined to evaluate all aspects of the system. Methods State-of-the-art in scenario-based and hash weather testing for AAD systems was analysed; based on the analysis, a team with diverse expertise in AAD development and testing defined a methodology for defining a set of harsh weather test cases. Results This paper proposes, and exemplifies the use of, a methodology to develop a representative set of test cases based on the defined operational design domain and use cases for an AAD system under development, considering the possibility of reproducing tests in different test environments with a focus on harsh weather. Conclusions We believe that our proposed methodology can accelerate the overall testing process and contribute to the difficult safety assurance challenges for automated vehicles.