Evolving Artificial Constructive Swarms - Experimental Models and Methodologies

This book briefly introduces to the bio-inspired concept of constructive swarms and focuses on the design and evolution of two different artificial constructive swarm models. First, a connectionist approach is taken in which the sensory and the actuator system of virtually flocking swarm individuals are connected by an artificial neural network. The swarms are evolved to satisfy the expectations of the breeder by means of manual evaluation. Exhaustive computing times turn out to be a great obstacle to achieving good results. A second model is designed to overcome this hurdle. As a result, a rule-based constructive swarm is automatically evolved to approximate pre-defined geometrical shapes. This book provides ideas and results to support research and applications in Computer Science, Arts and Architecture.