Employing agents to develop integrated urban models: Numerical results from residential mobility experiments

Multi-Agent systems are a powerful technique to analyse spatially distributed systems of heterogeneous autonomous actors with bounded information and computing capacity who interact locally. A review of recent urban models relying on multi-agent technology learns however that these models at best only start to explore this potential. In this paper, we present a model, simulating the process of residential mobility, fully exploiting the agent-potential, integrating behavioural concepts such as joint-decisions making, bounded rationality, pro-active reasoning, cognitive mapping, etc. We will discuss the conceptual framework, analyse some numerical results and make suggestions as to how to validate such an 'artificial-society' model.