Agent-based simulation of autonomous cars

More and more highly automated vehicles arrive in the consumer market and fully autonomous vehicles are predicted to become available in the next few years. Autonomous vehicles promise to fundamentally change mobility - for users as for planners. Transport models and simulations are required to prepare for these changes. This paper suggests agent-based transport models as a suited mean to model future transport scenarios including autonomous vehicles. The multi-agent transport model MATSim is presented in detail and some possible research questions on autonomous vehicles - the future car fleet size, future demand patterns, and the interaction between public transport and autonomous vehicles - are introduced. Reason is given, why agent-based models are particularly suited to investigate these questions.