Developing a Model to Map Global Positioning System (GPS) Data onto Transportation Networks

University Of Kentucky Research Foundation, Lexington KY

Investigators

Abstract

The United States' global positioning system (GPS), NAVSTAR, consists of a constellation of 24 satellites orbiting in high altitude. GPS receivers that can be placed in any type of vehicle, including automobiles, trucks, buses, airplanes, trains and ships, calculate their position on the earth based on their distance from four or more of these satellites. This ability to determine the location of vehicles in time has many useful applications in transportation including tracking vehicles on networks for navigation or travel surveys, as well as collecting and storing data about the networks of roads, waterways or even pipelines for facility inventories. Although development of a technique such as that proposed here to map GPS point data to linear network configurations has many applications, it is routing behavior and route choice as an improvement to traffic assignment models in urban planning that are specifically of interest to this researcher. The objective of this research is to develop a mathematical model to determine routes on networks from GPS point data in order to automate the route data collection process such that the recording of routes can become routine and widespread. The data points collected by GPS receivers for the route of a vehicle travelling in a transportation system are recorded as single points in space. Network segments such as roads, even though they have two dimensions including width, are represented as one dimensional lines in a GIS or other network database. Even if there were not errors in both the GPS points and the network data it would be necessary for the GPS points to be translated to links in order to consider the route behavior or optimize it in the network system. Currently, errors that can not be completely eliminated in both the GPS collected points as well as the base linear networks make the process of matching the GPS data onto the network complicated. In many cases, humans are needed to directly transform the GPS point data into a representation consisting of network line segments for analysis. This is not practical on a large scale for full transportation networks where a large number of vehicles are being tracked. In this project GPS point data for automobiles traveling on an urban road network will be collected (urban roads will be used because their close spacing as well as the interference from buildings challenges the methodology). Data will be collected during various weather conditions as well as at different times of the day when satellite configuration varies and affects measurement quality. The known actual routes will be manually entered into the road network database and a mathematical model will be developed to transform the GPS points into complete routes in link by link network (line) nomenclature. The model will make use of a minimum path algorithm and link impedances defined based on GPS point density, velocity and bearings. Goodness of fit measures that determine the extent of match between the actual and modeled routes will be used to evaluate alternative model specifications. This award is made under the Exploratory Research on Engineering the Transport Industries (ETI) program solicitation.