Verifying heterogeneous multi-agent programs

We present a new approach to verifying heterogeneous multi-agent programs — multi-agent systems in which the agents are implemented in different (BDI-based) agent programming languages. Our approach is based on meta-APL, a BDI-based agent programming language that allows both an agent’s plans and its deliberation strategy to be encoded as part of the agent program. The agent programs comprising a heterogeneous multi-agent program are first translated into meta-APL, and the resulting system is then verified using the Maude term rewriting system. We prove correctness of translations of Jason and 3APL programs and deliberation strategies into meta-APL. Preliminary experimental results indicate that our approach can significantly out-perform previous approaches to verification of heterogeneous multi-agent programs