Vehicle trajectory modelling with consideration of distant neighbouring dependencies for destination prediction

Abstract

Vehicle trajectory modelling is an essential foundation for urban intelligent services. In this paper, a novel method, Distant Neighbouring Dependencies (DND) model, has been proposed to transform vehicle trajectories into fixed-length vectors which are then applied to predict the final destination. This paper defines the problem of neighbouring and distant dependencies for the first time, and then puts forward a way to learn and memorize these two kinds of dependencies. Next, a destination prediction model is given based on the DND model. Finally, the proposed method is tested on real taxi trajectory datasets. Results show that our method can capture neighbouring and distant dependencies, and achieves a mean error of 1.08 km, which outperforms other existing models in destination prediction significantly.

Vehicle Destination Prediction Using Bidirectional LSTM with Attention Mechanism

“…However, basic RNNs are limited in dealing with longer sequence data due to the exploding and gradient vanishing phenomena [17,23]. The introduction of gate structures, found in LSTMs, has shown to partially mitigate these problems, and has become widely adopted in both trajectory [24][25][26] and destination prediction problems [17,19,[27][28][29], achieving promising outcomes. Brébisson et al [17] concluded that using a Bidirectional LSTM (BiLSTM) improved prediction performance as compared to unidirectional LSTM, due to the inputs being accessed in both directions.…”

Quantifying the Tradeoff Between Cybersecurity and Location Privacy

“…Deep neural networks have been gaining attention from both academia and industry to deal with GPS trajectory data for prediction tasks such as vehicle or trajectory classification [25], [26], travel prediction [27], [28], [29], and characterizing driving styles [30]. For trajectory data of vehicle movement, although some research suggests that extracting semantic and statistical features can improve the performance (e.g., the accuracy of detection and ROC, etc.)…”

“…This dataset includes trajectories from 442 taxis in Porto, Portugal between July 1st, 2013 and June 3rd, 2014. It was released in Kaggle ECML/PKDD 15: Taxi Trajectory Prediction (I) competition and has been widely used in academia for benchmarking new algorithms for trajectory modeling [29], [31], [36] and malicious trip detections [13].…”

“…In our analyses, we use the Long-Short Term Memory (LSTM) model [40], which is a modified version of RNNs. The reason is that the former is capable of learning long-term spatial dependencies between GPS points that are far apart from each other [29], namely, those GPS points that are more likely to co-occur in vehicle trajectories.…”