Abstract Simulation Scenario Generation for Autonomous Vehicle Verification

Abstract

Autonomous vehicle technology has the potential to have a significant impact on the transportation system and urban life. However, the autonomous vehicles must be proven to be at least as safe as human-driven vehicles before they are accepted as a new mode of transportation. Current methods of autonomous vehicle verification such as shadow driving or annotated images based testing are costly, slow and resource intensive. Hence, modeling and simulation is an indispensable asset to achieve verification goals for autonomous vehicles. In this paper, we propose an abstract simulation scenario generation framework for autonomous vehicle verification. The scenes and the related assertions are defined by a matrixbased semantic language and translated into test scenarios in simulation. The framework allows the design of all possible road topologies and the validation of generated scenarios. The scenarios generated in the framework form a ground truth for possible extended tests of rare conditions in other platforms.