Synthetic Fuzzy Evaluation Method of Trajectory Similarity in Map-Matching

Abstract

Despite the rapid development of concepts used in current map-matching algorithms, the continuous movement of a vehicle is often ignored or just limited to 1 or several previous matches. This may result in errors in complicated situations such as at Y-junctions, and entrance or exit ramps of a highway. The trajectory-based map-matching algorithm, which determines the matching road by comparing the vehicle trajectory against candidate roads, has the potential to overcome this limitation. However, the curve-to-curve matching as a simple form of trajectory-based map-matching fails to address this. The key issue of a trajectory-based map-matching algorithm is how to evaluate the similarity between the trajectory and the possible traveling roads. This article develops a synthetic fuzzy evaluation method to address the issue. The similarities are quantified in terms of location, shape, direction, and behavior. As a new input in map-matching, vehicle behavior refers to changes in the motion of the vehicle such as turning, which is restrained by the geometry of traveling road. The article also proposes a multilevel fuzzy synthetic evaluation process to assess the similarities and hierarchically synthesize them into a final evaluation. The method is evaluated using GPS positions recorded with a vehicle traveling in the urban area of Beijing, China. The traveled road paths include complicated roads conditions such as flyovers, highway entrances and exits. The method identifies more than 98% of the road segments correctly, showing a significant improvement over existing map-matching algorithms.