Skip to Main Content
Testing can be functional or structural. Neglecting one technique to the prejudice of the other is fundamentally flawed. In this paper, we propose an automated test case generation for agent based systems. The approach uses the system's specification and implementation. The functional testing uses a combination of the formal language Maude and a new kind of diagrams called InterElement Requirement Diagrams (IRD). For the structural testing, we propose a coverage criterion where the generated test cases will provide a level of coverage in the system's execution. We achieve our goal by translating agent's plans to Petri Nets where the computation is performed in the places and decisions, posted goals or dispatched messages are performed in the transitions. We have three levels of criteria: place, transition and plan.