Deducing mode and purpose from GPS data

Abstract

Over the past several years, there has been increasing interest in the use of Global Position System (GPS) devices as a means to measure people’s travel. These devices have been used quite widely to check the validity of self-reports of travel through travel diaries and similar conventional travel survey procedures. With the development of lightweight, high sensitivity GPS devices, there has been increasing interest in their use. GPS devices can collect and record data on points during a person’s travel, providing the latitude, longitude, (and altitude, if desired), the time, the speed of motion, and the heading at each such point, along with data quality measures. Data can be collected as frequently as every second, thereby providing a clear documentation of the path travelled. GPS devices cannot, however, collect data on the mode or the purpose of travel, both of which are essential components of the data record required for transport planning purposes. In this paper, we describe work that we have recently undertaken in which we have developed a set of heuristic rules for determining each of the mode of travel and the purpose of the trip, albeit with the use of comprehensive GIS data bases for the areas where the GPS surveys have been conducted. In the case of mode of travel, it is necessary to have a GIS that includes not only the street system, but also the bus routes and any rail lines or other modes of travel (such as ferry in the case of Sydney), where those modes do not travel along the street system. The rules we have developed consider the average speeds, the maximum and minimum speeds, and also use certain procedures to eliminate potentially spurious spot speed information. They also use information about the transport network and the availability of bicycles and cars to the survey participants. In the case of trip purpose, it is necessary to have a GIS of the land use of all parcels of land in the urban area for the survey, and we also ask people to tell us the addresses of all workplaces of members of the household, the school addresses for schools attended by household members, and the two most frequently used grocery stores or supermarkets. The rules then use the length of stay at the location and whether or not the location is one of the workplaces, schools, or stores identified. Frequency of visits during one or more weeks of data recording is also used to help identify the purpose of the trip. For trips that have one or both ends at a location other than home or one of the given addresses, the purpose is deduced from the land use information and the duration and frequency of the visit.