My task was to design and implement a multi-agent system capable of traffic modelling. The aim of the task was to create an agent-community for three goals: modelling traffic situations with the definition of the road network and the participating drivers’ behaviours, generating simulated traffic data and experimenting with changes in the network and in the behaviours.
I have designed the basic components of the system: the agents responsible for the intersections, roads and the cars travelling on them. Additionally, a synchronization-agent has been designed. A scheme for simulation of different traffic conditions has been laid for use in the system. I designed a communication protocol between the agents, and the composing messages that created the simulation itself. Three degrees of freedom has been defined in the decisions of the drivers: calculation of speed when travelling on a road, whether or not to give the right-of way in an intersection and the planning of the course.
The planned agents have been implemented in Eclipse-EJADE environment. A support framework has been constructed for handling the descriptions of traffic situation, their road network and participating drivers. This framework is also capable of automated simulation and data collection.
With the complete system three different simulations has been done with the aim of revealing the differences between the driver behaviours, and analyzing the traffic data of the road network. The outcome of these simulations has been evaluated by different measures, which led to conclusions about the effectiveness of the driver strategies, and improvements in the traffic conditions of the road network.