Multiple agent autonomous hybrid control systems
Wolf Kohn, Anil Nerode · 2005
Current research efforts in the development of a formal model for the control of dynamic, real-time autonomous systems in which multiple decision makers, control the plant are surveyed. The model provides a formal framework for expressing the interaction between evolution (continuous) and knowledge (discrete) components. This is the central characteristic of hybrid systems. The essence of the interaction between the evolution and discrete components is the continuity of the behavior of the system, as viewed by each of its agents, with respect to the topology defined by the knowledge. The model is being developed via a computational architecture whose central functionality is online, distributed theorem proving and reactivity. The central elements of this architecture are the controlling agents and the interagent communication network. Each agent is provided with an inferencer, a theorem planner, and a communication protocol device.>