| Multi-Agent System is a significant research branch of distributed artificial intelligence. The thesis takes the RoboCup-Rescue simulation system, a typical MAS, as research background and conducts the correlation research on task control mechanism in process of robot rescue simulation. Because of the heterogeneous characteristic of agents, complex relation between tasks and the dynamic environment, the thesis divide the task control problem to task allocation and task coordination, and the two corresponding algorithms are put forward.A Multi-Agent task allocation algorithm based on auction is proposed to solve the task coordination problem of the homogeneous agents. By introducing the auction algorithm a nearly optimized task allocation scheme can be inferred after a comprehensive evaluation to the benefit and cost of the task be evaluated comprehensively. The dynamic environment may cause a lag of task allocation scheme, for adjusting the scheme dynamically and in real-time the Token-Ring be introduced for controlling the communication in the adjustment process and relief the burden of communication.A Multi-Agent task coordination algorithm based on functional constraint graph is put forward to solve the task coordination problem of the heterogeneous agents. Functional constraint graph is composed of the time constraint graph and the function weight. The time constraint graph is used to describe the dependent and restraint relationship among the tasks of heterogeneous agents. The function weight is proposed to indicate every task's contribution to the overall goal of MAS. The operation of Functional constrain function can return the importance degrees of a series of tasks in functional constrain graph, and they are the basis of how to select the coordination scheme.The task control algorithm proposed in this thesis has been successfully adopted in a RoboCup-Rescue simulation team CSU_YunLu. The team's achievement proved that the algorithm is practicable and effective. |
PDF Full Text Request