Distributed intelligent control of hybrid multiagent systems

This dissertation focuses on the problem of the coordinated control of multiple distributed systems-of-systems in which each system exhibits both continuous-valued and discrete-event dynamic characteristics. Real-life examples of such multiple “hybrid” systems include teams of U&barbelow;nmanned A&barbelow;ir V&barbelow;ehicles (UAVs), V&barbelow;ertical/S&barbelow;hort T&barbelow;ake-O&barbelow;ff and L&barbelow;anding (V/STOL) aerospace vehicles; communication networks and certain interconnected electric power systems.; We develop a new framework for the integrated modelling; control and coordination of such multiple multi-mode dynamical systems. It is based on a Systems- and Control-oriented enhancement to the “Intelligent Agents” paradigm which we refer to as the H&barbelow;ybrid I&barbelow;ntelligent C&barbelow;ontrol A&barbelow;gent (HICA). Practically speaking, the HICA is an intelligent agent which has an embedded hybrid control system at its structural core. In this framework, a form of knowledge-based deliberative planning is integrated with a set of (near real-time) verified hybrid control primitives and coordination logic to provide full-authority coordinated control of systems of agents. A fundamental advantage of this scheme is that agent systems can be synthesized in a hierarchically consistent bottom-up manner from a single agent to supervisory teams of goal-directed multiagent systems which are amenable to mathematical description.; Several useful results that relate to the structure of multiagent HICA, single-agent and multi-agent stability are then developed. These enable a precise characterization of internal stability (for a single agent) and multiagent stability (for teams of agent) as well as the development of sufficient conditions for the former to hold. A framework is also developed to study the multiagent stability of agents interconnected by lossy communication media.; In addition to several illustrative examples throughout this thesis, the HICA methodology for developing, controlling and coordinating systems of agents, are demonstrated in experimental and simulation scenarios which involve autonomous unmanned air and ground vehicle teams.