Method for investigating intradriver heterogeneity using vehicle trajectory data: A Dynamic Time Warping approach

Taylor, Jeffrey; Zhou, Xuesong; Rouphail, Nagui M.; Porter, Richard J.

doi:10.1016/j.trb.2014.12.009

Cited by 101 publications

(45 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…where p (1) k ∈ R 2 and p (2) k ∈ R 2 are the position of two vehicles in the driving encounter at discrete time t k (Fig. 3), K ∈ N is the length of the driving encounter x.…”

Section: A Driving Encountermentioning

confidence: 99%

Clustering of Driving Encounter Scenarios Using Connected Vehicle Trajectories

Wang

Ramesh

Zhu

et al. 2020

IEEE Trans. Intell. Veh.

View full text Add to dashboard Cite

Multi-vehicle interaction behavior classification and analysis offer in-depth knowledge to make an efficient decision for autonomous vehicles. This paper aims to cluster a wide range of driving encounter scenarios based only on multi-vehicle GPS trajectories. Towards this end, we propose a generic unsupervised learning framework comprising two layers: feature representation layer and clustering layer. In the layer of feature representation, we combine the deep autoencoders with a distancebased measure to map the sequential observations of driving encounters into a computationally tractable space that allows quantifying the spatiotemporal interaction characteristics of two vehicles. The clustering algorithm is then applied to the extracted representations to gather homogeneous driving encounters into groups. Our proposed generic framework is then evaluated using 2,568 naturalistic driving encounters. Experimental results demonstrate that our proposed generic framework incorporated with unsupervised learning can cluster multi-trajectory data into distinct groups. These clustering results could benefit decisionmaking policy analysis and design for autonomous vehicles.Index Terms-Multi-vehicle behavior clustering, autoencoder, vehicle trajectory, unsupervised learning.

show abstract

“…where p (1) k ∈ R 2 and p (2) k ∈ R 2 are the position of two vehicles in the driving encounter at discrete time t k (Fig. 3), K ∈ N is the length of the driving encounter x.…”

Section: A Driving Encountermentioning

confidence: 99%

Clustering of Driving Encounter Scenarios Using Connected Vehicle Trajectories

Wang

Ramesh

Zhu

et al. 2020

IEEE Trans. Intell. Veh.

View full text Add to dashboard Cite

show abstract

“…This distance matrix is called the "local cost matrix" for the alignment of the sequences and [59]. Then, we calculate the cumulative cost matrix ( ), an × matrix that stores the cumulative smallest cost required to arrive at each element in the matrix by following an express command from (1, 1) to ( , ) [60,61]. Each element of is calculated using (1), which also identifies the search direction.…”

Section: Conventional Dtw Algorithm and Its Application Tomentioning

confidence: 99%

New Approach for Snow Cover Detection through Spectral Pattern Recognition with MODIS Data

et al. 2017

View full text Add to dashboard Cite

Snow cover plays an important role in climate and hydrology, at both global and regional scales. Most previous studies have used static threshold techniques to detect snow cover, which can lead to errors such as misclassification of snow and clouds, because the reflectance of snow cover exhibits variability and is affected by several factors. Therefore, we present a simple new algorithm for mapping snow cover from Moderate Resolution Imaging Spectroradiometer (MODIS) data using dynamic wavelength warping (DWW), which is based on dynamic time warping (DTW). DTW is a pattern recognition technique that is widely used in various fields such as human action recognition, anomaly detection, and clustering. Before performing DWW, we constructed 49 snow reflectance spectral libraries as reference data for various solar zenith angle and digital elevation model conditions using approximately 1.6 million sampled data. To verify the algorithm, we compared our results with the MODIS swath snow cover product (MOD10 L2). Producer's accuracy, user's accuracy, and overall accuracy values were 92.92%, 78.41%, and 92.24%, respectively, indicating good overall classification accuracy. The proposed algorithm is more useful for discriminating between snow cover and clouds than threshold techniques in some areas, such as those with a high viewing zenith angle.

show abstract

“…Intra-driver variability accounts for a significant proportion of deviations between simulated and empirical trajectories (Kesting and Treiber, 2009). We have developed enhanced Dynamic Time Warping (DTW) for calibrating synthetic trajectories from Newell's car following model that can reflect intra-driver heterogeneity (Taylor et al, 2015). Similar methods have been used successfully to generate link driving schedules compatible to be used with MOVES 2010 (Aziz and Ukkusuri, 2015).…”

Section: Day-to-day Emission Dynamics In a Work Zone Test Casementioning

confidence: 99%