PREDICTION AND DISSEMINATION SYSTEM FOR TRAVEL TIME UTILIZING VEHICLE DETECTORS

Abstract

Heavy traffic congestion and an increasing number of traffic accidents in the central city of Aomori, Japan, led to the development of a system that provides drivers with the predicted travel time along the congested route in peak hours. An outline of the system is that a traffic information central processing unit installed at the Aomori Traffic Control Center analyzes past traffic data, such as average volume, time-occupancy, and speed of vehicles. These data are uploaded from vehicle detectors on the route specified to offer travel time information. The central computer also predicts a travel time drivers would probably take during about 60 minutes in the immediate future. The system employs two kinds of prediction models: one is an auto regression model that makes predictions based on the change of traffic condition during the latest 30 minutes, and the other is a statistical model that also predicts based upon the difference from travel times in the previous several days. The system has been applied to two trunk roads in Aomori. Travel time offering media are roadside variable message signs. A local radio network has often broadcast the travel time information as well. The system has been well received by the public. A reason for this may be that the average error in the predicted travel time is being limited within 3 minutes. Also, the system is effective in eliminating traffic congestion.