Beyond the Ad-hoc and the Impractically Formal: Lessons from the Implementation of Formalisms of Intention

摘要

This paper describes a BDI-inspired abstraction layer that is implemented in Soar to increase the scalability and flexibility of knowledge engineering in Soar-based agent systems. We motivate the BDI abstraction with a specific example of a multi-agent system that is being developed to explore human command and control of heterogeneous robotic entities. We then describe how we have implemented deontics and communication aspects of Joint Intentions via the abstraction layer and illustrate the strengths, limitations, and lessons learned in this approach.