A model for traffic simulation and a simulation language for the general transportation problem

Abstract

Widespread interest in better traffic control techniques has resulted in the past few years because of the rapidly increasing numbers of vehicles in American cities. This paper describes a modeling technique which promises to be useful for traffic simulation and, thereby, aiding development of more advanced on-line signal control systems. For this method, simulation is accomplished by stepping individual vehicles through a traffic network system in accordance with driver responses to changing traffic and environmental conditions. During each time increment, a driver's response such as speed up, slow down, etc., is computed from various input conditions normally available to a driver. These inputs might include "is vehicle ahead slowing down," "is vehicle ahead stopped," etc. The vehicle's new positions, velocities, etc., are then computed and the next vehicle considered. Thus, the driver's responses to various input conditions are preprogrammed for the various driver types and intersection conditions. Vehicle arrival rates, driver characteristics, and vehicle flow paths may be functions of various statistical distributions. However, by also allowing preprogrammed driver actions, a desirable mixture between statistical and heuristic operations is obtained. Since the simulation can be used to predict traffic conditions, on-line simulation can then possibly be used as criteria for the selections of traffic signal control patterns or other control techniques.