Improving the Performance of a Topological Map-Matching Algorithm Through Error Detection and Correction

Abstract

Map-matching algorithms integrate data from positioning sensors with a digital map in order, first, to identify the road link on which a vehicle is traveling, and second, to determine the vehicle's location on that link. Due to errors in positioning sensors, digital maps, and the map-matching (MM) process, MM algorithms sometimes fail to identify the correct road segment from the candidate segments. This phenomenon is known as mismatching. Identification of the wrong road link may mislead users and degrade the performance of a location-based intelligent transportation system (ITS) and services. The main objective of this article is to improve a topological map-matching (tMM) algorithm by error detection, correction, and performance re-evaluation. Errors in a tMM algorithm were determined using data comprising 62,887 positioning points collected in three different countries (the United Kingdom, the United States, and India). After map-matching, each mismatched case was examined to identify the primary causes of the mismatches. A number of strategies were developed and applied to reduce the risk of mismatching thus enhancing the tMM algorithm. An independent data set of 5,256 positioning points collected in and around Nottingham, UK, was employed to re-evaluate the performance of the enhanced tMM algorithm. The original tMM algorithm correctly identified the vehicle's position 96.5% of the time; after enhancement this increased to 97.8%. This compares very well with the performance of tMM algorithms reported in the literature. The enhanced tMM algorithm developed in this research is simple, fast, efficient, and easy to implement. Since the accuracy offered by the enhanced algorithm is found to be high, the developed algorithm has potential to be implemented in real-time location-based ITS applications.