The complexity of the driving behavior restricts the realism of traffic simulation. This paper proposed that vehicle mobility models should be established according to diverse driving styles to further approximation of real driving behavior. With Krauss model represented, the conservative (driving style) of safe distance car-following model is analyzed. The analysis means that real vehicles can occasionally break the safe distance rule, and on average, real vehicle gap is slightly smaller than that in the Krauss model. An aggressive car-following model is proposed in the view of driving style. Simulation results show the new model can simulate aggressive driving style, which has significance to simulate traffic using diverse driving style models. Since it breaks the safe distance rule, the new model has the possibility of generating rear-end collisions when simulating. Drivers’ characteristics, prediction behavior, the cause of accidents, and the effects of time granularity on a simulation are studied. The concept of “road black hole” is put forward, which is believed to reduce velocity of traffic flow.
An FPGA-Based Neuro-Fuzzy Sensor for Personalized Driving Assistance
