Abstract
Facing the economics globalization and the trend of job specialization in industry, modern enterprises start to separate and outsource their product development tasks in order to strengthen core competence and maintain competitiveness. This research investigates multi-agent based design negotiation methods in a distributed environment with limited access to information. The focus is on improving centralized “price based negotiation” and de-centralized “asynchronous backtracking” methods. We first propose a novel conflict resolving mechanism that adjusts individual negotiation policies each time when agreement cannot be achieved during the negotiation process. The price based negotiation approach is next extended to distributed decision making within a multi-stage hierarchy. These enhancements provide more flexible negotiation and better efficiency compared to previous work. This work also implements the proposed methods with multiple agent system (MAS) technologies to facilitate the negotiation process by automating lengthy iterations. Finally, we apply the methods to different distributed design collaboration situations including tolerance allocation, resource allocation among multi manufacturing sites, reliability distribution in complex systems development, and LOD data transmission in 3D streaming. The test results not only validate this work, but they also demonstrate the practicality of the agent-based distributed negotiation methods.