Towards world model-based test generation in autonomous systems

This paper describes a model-based test generation approach for testing autonomous systems interacting with their environment (i.e., world). Unlike other approaches that assume a static world with attributes and values, we present and test the world dynamically. We build the world model in two steps: a structural model that constructs environmental factors (i.e., actors) and a behavioral model that describes actors' behaviors over a certain landscape (i.e., snippet). Abstract world behavioral test cases (AWBTCs) are then generated by covering the behavioral model using graph coverage criteria. The world model-based test generation technique (WMBTG) is used on an autonomous ground vehicle (AGV).