TY - JOUR
T1 - Offline map matching using time-expanded graph for low-frequency data
AU - Tanaka, Akira
AU - Tateiwa, Nariaki
AU - Hata, Nozomi
AU - Yoshida, Akihiro
AU - Wakamatsu, Takashi
AU - Osafune, Shota
AU - Fujisawa, Katsuki
N1 - Funding Information:
This research project was supported by the Japan Science and Technology Agency (JST), the Core Research of Evolutionary Science and Technology (CREST), and JSPS KAKENHI Grant No. JP 16H01707 and 21H04599.
Publisher Copyright:
© 2021 Elsevier Ltd
PY - 2021/9
Y1 - 2021/9
N2 - Map matching is an essential preprocessing step for most trajectory-based intelligent transport system services. Due to device capability constraints and the lack of a high-performance model, map matching for low-sampling-rate trajectories is of particular interest. Therefore, we developed a time-expanded graph matching (TEG-matching) that has three advantages (1) high speed and accuracy, as it is robust for spatial measurement error and a pause such as at traffic lights; (2) being parameter-free, that is, our algorithm has no predetermined hyperparameters; and (3) only requiring ordered locations for map matching. Given a set of low-frequency GPS data, we construct a time-expanded graph (TEG) whose path from source to sink represents a candidate route. We find the shortest path on TEG to obtain the matching route with a small area between the vehicle trajectory. Additionally, we introduce two general speedup techniques (most map matching methods can apply) bottom-up segmentation and fractional cascading. Numerical experiments with worldwide vehicle trajectories in a public dataset show that TEG-matching outperforms state-of-the-art algorithms in terms of accuracy and speed, and we verify the effectiveness of the two general speedup techniques.
AB - Map matching is an essential preprocessing step for most trajectory-based intelligent transport system services. Due to device capability constraints and the lack of a high-performance model, map matching for low-sampling-rate trajectories is of particular interest. Therefore, we developed a time-expanded graph matching (TEG-matching) that has three advantages (1) high speed and accuracy, as it is robust for spatial measurement error and a pause such as at traffic lights; (2) being parameter-free, that is, our algorithm has no predetermined hyperparameters; and (3) only requiring ordered locations for map matching. Given a set of low-frequency GPS data, we construct a time-expanded graph (TEG) whose path from source to sink represents a candidate route. We find the shortest path on TEG to obtain the matching route with a small area between the vehicle trajectory. Additionally, we introduce two general speedup techniques (most map matching methods can apply) bottom-up segmentation and fractional cascading. Numerical experiments with worldwide vehicle trajectories in a public dataset show that TEG-matching outperforms state-of-the-art algorithms in terms of accuracy and speed, and we verify the effectiveness of the two general speedup techniques.
UR - http://www.scopus.com/inward/record.url?scp=85110289491&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85110289491&partnerID=8YFLogxK
U2 - 10.1016/j.trc.2021.103265
DO - 10.1016/j.trc.2021.103265
M3 - Article
AN - SCOPUS:85110289491
VL - 130
JO - Transportation Research Part C: Emerging Technologies
JF - Transportation Research Part C: Emerging Technologies
SN - 0968-090X
M1 - 103265
ER -