Emergent Aerospace Designs Using Negotiating Autonomous Agents

the design cycle is either extremely long or the designers have to rely on approximations. This paper presents a distributed design methodology The use of multi-disciplinary knowledge at the product where designs emerge as a result of the negotiations bedevelopment stage presents opportunities for improving the tween different stake holders in the process, such as cost, product cost to performance ratios. However, this desired performance, reliability, etc. The proposed methodology capability, along with the changing customer requirements, uses autonomous agents to represent design decision makplaces a heavy burden on the design decision makers who ers. Each agent influences specific design parameters in must constantly re-engineer products to comply with the order to maximize their utility. Since the design paramecustomer needs. In this environment it is critical to have ters depend on the aggregate demand of all the agents in a set of integrated tools that can efficiently handle the rethe system, design agents need to negotiate with others in current design cycles and at the same time manage the the market economy in order to reach an acceptable utility complex relations between the different design criteria. value. This paper addresses several interesting research isThis paper presents a Multi-Agent Design Architecture sues related to distributed design architectures. First, we (MADA) which incorporates both the needs by providing present a flexible framework which facilitates decomposia computational environment suitable for evaluating nution of the design problem. Second, we present overview of merous design scenarios and a distributed solution search a market mechanism for generating acceptable design conmethod that can handle multiple design criteria. The disfigurations. Finally, we integrate learning mechanisms in tributed search methodology uses negotiations between difthe design process to reduce the computational overhead. ferent stake holders, or design decision makers, to guide the search in the design parameter space. An important