Comparing Proving Ground and Simulation Testing for Autonomous Car Functionality

Abstract

Safety and reliability are the most critical attributes of vehicles when it comes to licensing. To ensure this two attributes, every vehicle has to undergo testing processes to ensure the driver's safety and the driving system's reliability. A significant factor in testing a driving system is the amount of kilometres a specific component has been tested on. These test drives can be automated, theoretically. This paper shows methodical approaches to investigate which aspects of a simulation are already feasible enough to be used in automated testing. Furthermore, it discusses how systems under test have to be designed to fit into an automated testing pipeline. It also analyses how validation by simulation performs economically in comparison to manual test drives.