Alan Garvey, Keith Decker, and Victor Lesser
Our real-time problem solver consists of a real-time scheduling component and a decision making component. These components are semi-autonomous and have a complex, bidirectional interface. We have found it useful to model this interface as a negotiation process, where conflicts between what each component wants to do are resolved through negotiation. A distinguishing feature of this work is that we argue that negotiation is useful, even within a single agent architecture.