Creating a core-enabler for evaluating scenarios of mixed vehicular traffic

Research Project, 2017

In the coming 30-50 years, we will see a mixed vehicular traffic with variable composition of e.g., human-driven vehicles, self-driving vehicles, electric vehicles, etc. at different times at different cities/areas. A smart city needs to start evaluating the different scenarios of a mixed vehicular traffic, for example to predict the crash rate and the efficiency rate of such traffic system, to be able to come up with a city design that supports the transition phase and eventually the future transportation. A traffic simulator is an essential tool for such evaluation and/or prediction. However, currently traffic simulators are fed with hypothetical or simple road user behaviours that do not fully reflect the real behaviour of the different traffic elements. In this project, we will extract distribution of parameters (e.g., speed, time headway) in different traffic environments from naturalistic driving data to make traffic simulator more realistic and thereby increase its usefulness. A realistic traffic simulator could also be used to help in designing autonomous driving systems, e.g., via estimating the crash risk of a mixed traffic system when certain driving styles are applied to autonomous driving systems.