Analysis of Traffic Processing Efficiency Based on Cellular Automaton Model

Abstract

To evaluate the traffic on a straight road, we analyze the driving behavior of self-driving cars and human-driving cars from the micro perspective. Based on cellular automaton model, we simulate two sorts of cellular by using two kinds of cars. Moreover, basic driving rules and peculiar lane-changing rules of these cars are designed. According to a one-day count of traffic on one route in Washington, the mixed distribution function with the method of least square estimate can be obtained. We simulate the incoming of vehicles then the cellular generation is acquired. Similarly, we simulate the outgoing of vehicles and gain the cellular output. Output-to-generation ratio is defined as the traffic processing efficiency, which is the most essential index in evaluating the traffic capacity. Two traffic count’s tipping points (high, low) merely related to the number of lanes could be calculated. The traffic processing efficiency is deteriorated sharply when the traffic count outnumbers the high tipping point. On the contrary, the traffic processing efficiency is improved to 100% when the traffic count is inferior to the low tipping point. The equilibria for the percentage of self-driving cars exists when the traffic count ranges between the two tipping points. Then we draw a conclusion that there should not be lanes dedicated to self-driving cars after analyzing the traffic processing efficiency and the critical value of different number of lanes.