This chapter presents the results of a decision analysis study on route choice by car drivers, which was designed to estimate parameters for a route selection algorithm for navigation systems. The author argues that a good route selection algorithm must result in leading car drivers over routes that they would have chosen if they had complete knowledge of their characteristics. Data from 49 volunteer subjects in Groningen in The Netherlands were analyzed. The subjects had different gender, age, education level, trip motive and driving experience. Four routes to a chosen destination were selected by each subject, who was given a large map of the city. The subject was asked to describe circumstances under which each route might be used. The subjects used the global preference method of selection. This gave a criterion variable for the correctness of the estimated weights used in selection. When the weights were used for a route selection algorithm, the following attributes seemed to result in selection of routes which would have been the first choice for 82% of the drivers: (1) waiting time for giving right of way; (2) waiting time for traffic lights; (3) road type; and (4) distance. Only 14% of the selected routes were second choices, only 4% were second worst, and none were worst. Route selection was similar for all types of drivers.
Abstract