Measuring Lane-changing Trajectories by Employing Context-based Modified Dynamic Time Warping
Hamedi, Hamidreza LU ; Shad, Rouzbeh and Jamali, Sadegh LU
(2023)
In Expert Systems with Applications
216.
- Abstract
- The spatial lane-changing (LC) behavior should be analyzed for the vehicles in transportation systems in order to identify the patterns of vehicles’ movements using the similarities detected in their lane-changing trajectories. The trajectory of an LC vehicle is a function of its context. The present paper utilized spatial footprints and external/internal contexts to contextualize a measure applicable to the similarities found between LC trajectories. While only the external context of the previous investigation was constrained to the surrounding vehicles on the road, this study has investigated the idea of the contribution of solar radiation to the lane-changing trajectory patterns. The similarities found between multi-dimensional... (More)
- The spatial lane-changing (LC) behavior should be analyzed for the vehicles in transportation systems in order to identify the patterns of vehicles’ movements using the similarities detected in their lane-changing trajectories. The trajectory of an LC vehicle is a function of its context. The present paper utilized spatial footprints and external/internal contexts to contextualize a measure applicable to the similarities found between LC trajectories. While only the external context of the previous investigation was constrained to the surrounding vehicles on the road, this study has investigated the idea of the contribution of solar radiation to the lane-changing trajectory patterns. The similarities found between multi-dimensional trajectories were determined by offering context-based modified dynamic time warping (CMDTW), and the CMDTW technique with the Next Generation Simulation (NGSIM) dataset was carefully analyzed. The weighting framework used for each dimension made it possible to quantify the similarities between lane-changing trajectories using the AHP technique. The obtained results showed that not only the lane-changing procedure depends on the conditions of the lane changer, but this procedure also depends on the solar radiation and the surrounding vehicles offered as the external contexts. Additionally, by including different dimensions, both internal and external contexts, the similarity results of LC trajectories turn into a more realistic phenomenon. The potential of the context-based modified dynamic time warping algorithm in the detection of a trajectory with the maximum similarity is also enhanced. Furthermore, in order to determine the LC trajectories, we used the Fuzzy C-means (FCM) clustering technique. We utilized Cohen’s kappa for the evaluation of the Fuzzy C-means results, and since the calculated Kappa score exceeds 0.8, the clustering algorithm has an excellent performance. The results obtained by comparing the suggested technique with commonly used similarity measurement techniques indicated that the accuracy of the CMDTW technique outperforms other techniques in the detection of the lane-changing trajectory patterns. The suggested CMDTW method has therefore been effective in the identification of the patterns of lane-changing trajectory. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/575ead2b-da47-4fc2-bc0f-38b856dfb63c
- author
- Hamedi, Hamidreza
LU
; Shad, Rouzbeh
and Jamali, Sadegh
LU
- organization
- publishing date
- 2023-04-15
- type
- Contribution to journal
- publication status
- published
- subject
- in
- Expert Systems with Applications
- volume
- 216
- article number
- 119489
- publisher
- Elsevier
- external identifiers
-
- scopus:85145971799
- ISSN
- 0957-4174
- DOI
- 10.1016/j.eswa.2022.119489
- language
- English
- LU publication?
- yes
- id
- 575ead2b-da47-4fc2-bc0f-38b856dfb63c
- date added to LUP
- 2023-01-04 11:15:13
- date last changed
- 2023-02-16 14:48:15
@article{575ead2b-da47-4fc2-bc0f-38b856dfb63c,
abstract = {{The spatial lane-changing (LC) behavior should be analyzed for the vehicles in transportation systems in order to identify the patterns of vehicles’ movements using the similarities detected in their lane-changing trajectories. The trajectory of an LC vehicle is a function of its context. The present paper utilized spatial footprints and external/internal contexts to contextualize a measure applicable to the similarities found between LC trajectories. While only the external context of the previous investigation was constrained to the surrounding vehicles on the road, this study has investigated the idea of the contribution of solar radiation to the lane-changing trajectory patterns. The similarities found between multi-dimensional trajectories were determined by offering context-based modified dynamic time warping (CMDTW), and the CMDTW technique with the Next Generation Simulation (NGSIM) dataset was carefully analyzed. The weighting framework used for each dimension made it possible to quantify the similarities between lane-changing trajectories using the AHP technique. The obtained results showed that not only the lane-changing procedure depends on the conditions of the lane changer, but this procedure also depends on the solar radiation and the surrounding vehicles offered as the external contexts. Additionally, by including different dimensions, both internal and external contexts, the similarity results of LC trajectories turn into a more realistic phenomenon. The potential of the context-based modified dynamic time warping algorithm in the detection of a trajectory with the maximum similarity is also enhanced. Furthermore, in order to determine the LC trajectories, we used the Fuzzy C-means (FCM) clustering technique. We utilized Cohen’s kappa for the evaluation of the Fuzzy C-means results, and since the calculated Kappa score exceeds 0.8, the clustering algorithm has an excellent performance. The results obtained by comparing the suggested technique with commonly used similarity measurement techniques indicated that the accuracy of the CMDTW technique outperforms other techniques in the detection of the lane-changing trajectory patterns. The suggested CMDTW method has therefore been effective in the identification of the patterns of lane-changing trajectory.}},
author = {{Hamedi, Hamidreza and Shad, Rouzbeh and Jamali, Sadegh}},
issn = {{0957-4174}},
language = {{eng}},
month = {{04}},
publisher = {{Elsevier}},
series = {{Expert Systems with Applications}},
title = {{Measuring Lane-changing Trajectories by Employing Context-based Modified Dynamic Time Warping}},
url = {{http://dx.doi.org/10.1016/j.eswa.2022.119489}},
doi = {{10.1016/j.eswa.2022.119489}},
volume = {{216}},
year = {{2023}},
}