Future complex engineered systems must be adaptable to, and function in, unpredictable situations, such as deep space or ocean explorations, hazardous waste cleanups, and search-and-rescue missions. To increase system adaptability, various multi-agent system approaches have been developed. From a design point of view, a critical question in developing such systems is: how can system adaptability be designed into complex systems based only on local interactions between (many) simple cells or agents? Although explicit cooperative methods have been applied to answer this question, their limitation in scaling-up has been recognized. In this paper, we introduce a meta-interaction model that can be used as a design approach towards multi-agent complex systems. The approach parameterizes behavioral interactions extended from the Boids swarm intelligence model by introducing dynamical variables into the system. The goal of the meta-interaction model is to provide a mapping for the prediction of collective functionality from local interactions and for the indication of local interactions based on desired functionality. The proposed model is described in detail and a computer simulation based case study of search-and-surround is presented to demonstrate the effectiveness of our proposed approach to designing complex adaptive systems.

This content is only available via PDF.
You do not currently have access to this content.