Research On Path Planning, Robust Identification And Robust Control Of Cooperative Multi-Mobile Robot Systems

With modularity, maintainability, extendibility, fault tolerance and robustness, multi-agent systems (MASs) appear to be well suited to solve complex problems for which classical methods cannot provide good results. By dividing the whole problem into several easier sub-problems and assigning to each single-agent, a global solution is obtained by the interactions among those single-agents solving the sub-problems locally, and the interactions between those single-agents and the host which carries out the dividing and assigning. In MASs, although the ability of each single-agent is limited, it can be improved by the cooperation and coordination among those single-agents, which simultaneously improves the performance of the whole system. Consequently, multi-agent systems have better performance, shorter time for task execution and lower energy consumption than single-agent systems.Because of the inaccuracy of modelling and the influence of their internal perturbation and external disturbance from environment, unavoidably, multi-agent systems have uncertainties, leading to the necessity of study of robustness of MASs. In this dissertation, evolutionary computation, a representation of intelligent control, is combined with modern robust control to be a robust algorithm for multi-agent systems. Meanwhile, the theoretical frame is also built. As a typical system of MASs, a cooperative multi-mobile robot system is at the core of this dissertation. Path planning of the cooperative multi-mobile robot system, robust identification and robust control of each mobile robot for tracking the planned paths are studied in five chapters as follows:Firstly, definition of multi-agent systems, proposal of robustness study of MASs, status and existing problems in path planning of mobile robots, in nonlinear system identification and in nonlinear system robust control are discussed.Then, a novel cooperative coevolutionary adaptive genetic algorithm and a novel fixed-length decimal encoding mechanism for paths of each mobile robot are proposed for path planning of cooperative multi-mobile robot system. Main contributions in this dissertation are the following. First of all, obstacles in the mobile robots' environment are described as polygons. Then, vertexes of all polygons are randomly numbered in decimal. The paths are then encoded...