Automated analysis and verification of agent behavior

Abstract

Comprehending and analyzing agent behavior is an arduous task due to complexities in agent systems and sophistication of agent behaviors, in addition to the common difficulties with any complex software system. This paper presents an integrated approach for the analysis and verification of behaviors of agentbased systems. The approach is a result of collaboration between the Tracer Tool and the TTL Checker, which together automate the analysis and verification of agents in an implemented agent system with the aim of aiding the user in redesigning, debugging, and maintaining the software system. The Tracer Tool ensures that the user's comprehension of the system behavior is accurate and provides explanations of anomalous behavior, which can be detected as a failed behavioral property by the TTL Checker. The integrated approach has been applied successfully in a case study in the domain of Unmanned Aerial Vehicles.

Verifying Robocup Teams

“…In [7] a combination of UML statecharts and hybrid automata was proposed for modeling multi-agent systems, and the method was applied to the task of model checking agents of the RoboCup rescue simulation league. In [3] a trace based approach is used to study a complex agent scenario. This paper is organized as follows.…”