Cooperative control of a network of multi-vehicle unmanned systems.

摘要

Development of unmanned systems network is currently among one of the most important areas of activity and research with implications in variety of disciplines, such as communications, controls, and multi-vehicle systems. The main motivation for this interest can be traced back to practical applications wherein direct human involvement may not be possible due to environmental hazards or the extraordinary complexity of the tasks. This thesis seeks to develop, design, and analyze techniques and solutions that would ensure and guarantee the fundamental stringent requirements that are envisaged for these dynamical networks.In another methodology, the game theory approach is used to formulate the consensus seeking problem in a "more" cooperative framework. For this purpose, a team cost function is defined and a min-max problem is solved to obtain a cooperative optimal solution. It is shown that the results obtained yield lower cost values when compared to those obtained by using the optimal control technique. In game theory and optimal control approaches that are developed based on state decomposition, linear matrix inequalities are used to impose simultaneously the decentralized nature of the problem as well as the consensus constraints on the designed controllers. Moreover, performance and stability properties of the designed cooperative team is analyzed in presence of actuator anomalies corresponding to three types of faults. Steady state behavior of the team members are analyzed under faulty scenarios. Adaptability of the team members to the above unanticipated circumstances is demonstrated and verified. Finally, the assumption of having a fixed and undirected network topology is relaxed to address and solve a more realistic and practical situation. It is shown that the stability and consensus achievement of the network with a switching structure and leader assignment can still be achieved. Moreover, by introducing additional criteria, the desirable performance specifications of the team can still be ensured and guaranteed.In this thesis, the problem of team cooperation is solved by using synthesis-based approaches. The consensus problem is defined and solved for a team of agents having a general linear dynamical model. Stability of the team is guaranteed by using modified consensus algorithms that are achieved by minimizing a set of individual cost functions. An alternative approach for obtaining an optimal consensus algorithm is obtained by invoking a state decomposition methodology and by transforming the consensus seeking problem into a stabilization problem.