Traffic flow simulation validation

Abstract

Transportation issues are a major concern among ordinary citizens, as majority of the population travels between places daily. Singapore being one of the highly dense countries in the world, have to cope with the travel demand of its population. This increasing need for transportation is one of the factors that motivates researchers to study more about the traffic to get a better understanding about the phenomenon. Traffic simulator is a method widely used to learn and evaluate potential solutions by replicating the environment and the vehicle's behaviours as it allows researchers to test complex behaviour and scenarios in a risk-free environment. This study focus is on evaluating whether the traffic flow generated by simulation framework of CityMoS, a microscopic agent-based traffic simulator is similar with the real traffic observed in Singapore, through calibrating the "numberofAgents" parameters in CityMoS which defines the number of agents in the simulation in an attempt to replicate and resemble the traffic condition in Singapore. Overall, by calibrating only the “numberofAgents” parameter in CityMoS, it is able to resemble approximately 15% of the traffic condition in Singapore. There are still many areas which can be done to enhance it further which is not covered in the scope of this project. Finally, a discussion on the reasons affecting the matching percentage and the recommendation on how to improve CityMoS is given for future implementation.