Model checking agent programming languages

Abstract

In this paper we describe a verification system for multi-agent programs. This is the first comprehensive approach to the verification of programs developed using programming languages based on the BDI (belief-desire-intention) model of agency. In particular, we have developed a specific layer of abstraction, sitting between the underlying verification system and the agent programming language, that maps the semantics of agent programs into the relevant model-checking framework. Crucially, this abstraction layer is both flexible and extensible; not only can a variety of different agent programming languages be implemented and verified, but even heterogeneous multi-agent programs can be captured semantically. In addition to describing this layer, and the semantic mapping inherent within it, we describe how the underlying model-checker is driven and how agent properties are checked. We also present several examples showing how the system can be used. As this is the first system of its kind, it is relatively slow, so we also indicate further work that needs to be tackled to improve performance.

Progression and Verification of Situation Calculus Agents with Bounded Beliefs

“…Most work on verification of agent systems/programs uses propositional modal logics and model checking techniques [3,32]. These include [6,18,47], and [53], which all focus on model-checking of BDI programs. Model checking (and satisfiability) in these propositional modal logics is decidable.…”

A New Proof System to Verify GDT Agents

“…Moreover, most of these systems use model-checking [1,8,5]. The most recent and promising work in this area is surely the Agent Infrastructure Layer (AIL) [3] and its connection with AJPF (Agent JPF), an adaptation of the Java Path Finder model-checker. AIL is a kind of intermediate language that can be used to give a formal semantics to most BDI agent languages (such as AgentSpeak or MetateM).…”

Formal Methods for the Certification of Autonomous Unmanned Aircraft Systems

“…This is written using the autonomous agent language Gwendolen [10], and we show that it can be verified against the same Rules of the Air using the agent model checker AJPF [2,11]. We introduce more advanced autonomous behaviour into the UAS agent, and verify that this acts in accordance with the subset of the Rules of the Air.…”

“…We check the Gwendolen program against the same logical requirements, but as SPIN only checks PROMELA programs, we must use a different model checker called AJPF [2,11] to establish correctness of the Gwendolen program with respect to the logical properties.…”

“…We use AJPF (for Agent JPF), which works by providing a Java interface for BDI agent programming languages called the Agent Infrastructure Layer (AIL) [17]. Interpreters for agent programming languages are written using the AIL, and the resulting Java program can then be verified via AJPF [2,11]. AJPF is, in turn, built on JPF, the Java PathFinder model checker developed at NASA Ames Research Center [25,16].…”