Multi-agent system for simulation of team behaviour in product development.

Teams play a crucial role in many product development activities. However, teams – as complex socio-technical systems – are challenging to study in a real-world setting. Computational models and simulations offer easily controllable and cost-effective tools for supporting team studies, and prove especially valuable in studies on the sufficient conditions for the emergence of various team properties and behaviours. As a consequence, computational models offer a possible explanation of the observed phenomena and provide means for theory-testing and hypothesis generation. Thus, this work develops a computational model of team behaviour in product development with the aim of enabling computational studies of team learning and adaptation. The dissertation develops a multi-agent system capable of representing various aspects of product development teams. Building on the existing theories, empirical studies and current teamwork models, the thesis derives a detail theoretical model of a designer, its cognition, its tasks, and interactions with others. The thesis implements the theoretical model in a multi-agent system directed at studies of the emergence of team properties and behaviours – in particular, team learning and adaptation. Extensive testing of the derived computational model concludes the goal of obtaining a multi-purpose research tool for studies of product development teams and confirms the appropriateness of the chosen simulation technique – agent-based modelling – for the intended research goal.