| Summary: | 碩士 === 國立中正大學 === 電機工程研究所 === 88 === The architecture of a multi-agent system is a distributed artificial intelligent architecture. Using a multi-agent architecture, we can solve large and complex problems that are difficult for a single agent to solve. In a multi-agent system, agents work together in a cooperation or competition environment. Every agent has its individual skills to accomplish tasks that are required by users or other agents. A new agent may be added into the system to improve the performance of the system or to increase the task domain of the system. Cooperation is a very important characteristic in a multi-agent system. We can say that agents are cooperating when the performance of the multi-agent system increases because of the addition of a new agent or when agents take some action to avoid or to solve potential or actual conflicts. When agents want to cooperate with others, task allocation and the coordination of action are two of the main problems. Task allocation is the problem of assigning tasks and problem-solving resources to an agent. Coordination is the problem of coordinating the action of agents to avoid the conflicts between them and to improve the efficiency of the system.
In this thesis, we introduce and discuss the task allocation strategies and the coordination strategies. Then, we introduce a goal-based analysis process to analyze the system that we want to construct with a multi-agent architecture. After the analysis, we can understand the cooperation relationship among agents in our system and understand when the system has the conflicts of goals and public resources between agents. Next, we discuss the considerations about how can we decide to take which task allocation strategy after the analysis process. Finally, we take the Intelligent Human-Computer Interface of the intelligent wheelchair system for application.
|
|---|
