Application of GPS/GIS Based Travel Mode Detection Method for Energy Efficient Transportation Sector

Abstract

The planning of efficient transportation systems is a central topic of concern as it saves energy and contributes to economic development both at national and household level. The development and optimization of highway and transportation engineering models such as transportation networks, parking facilities and as well as reduction of traffic congestion etc., require information on travel modes. To overcome the burdens for the respondent and the other drawbacks of the conventional methods related to travel mode detection, combination of Global Positioning System (GPS) and Geographical Information System (GIS) were tested in this research. After validating the accuracy of the smartphone based ‘My Tracks’ GPS mobile application, GPS travel route with a manually entered travel mode diary was used to collect data from randomly selected days and students in two faculties in the University of Peradeniya. Out of 125 datasets, 99 accurate datasets were selected for the analysis using ‘ArcGIS’ software coupling with ‘STATA’ statistical software. Rule-Based algorithms and heuristic based methods are two main approaches that were used to detect four travel modes (Walk, Bus, Motor Vehicles and Motor Bicycles). Differentiation of Bus mode from the motor vehicles (except motor bicycles) was the most challenging part of the research and a country specific novel method was introduced in this research. Travel mode detection results were compared with relevant correctly completed travel mode diaries to validate the methodology. The overall travel mode detection accuracy of the new method is 96% for the accurate data sets compared to conventional travel mode diary method. The success rate of collecting an accurate data set is 79%. The GPS/GIS based travel mode detection could be applied in energy efficient development in the transportation sector. In future research, the bus mode detection method invented in this research could be used to detect driver behaviour.