Spatial Temporal Attention based Target Vehicle Trajectory Prediction for Internet of Vehicles

TL;DR

The paper tackles accurate vehicle trajectory prediction in IoV/ITS by introducing STATVTPred, a model that marries Graph Attention Networks for local road-network spatial features with Transformer-based temporal modeling, all aligned with a map-matched road-sequence representation $G=(V,E)$. By preprocessing GPS data, applying UBODT-based map data, and enforcing connectivity via a Filter layer, the approach yields continuous, realistic trajectory predictions. Empirical results on Beijing and Chengdu taxi datasets show substantial AMR gains over LSTM Encoder-Decoder and Transformer baselines (e.g., AMR of $73.07\%$ and $78.93\%$ respectively), with strong ablation evidence for the spatial and filtering components. This work advances ITS capabilities in autonomous navigation, predictive routing, and traffic management by providing a robust, graph-informed, spatio-temporal trajectory predictor.

Abstract

Forecasting vehicle behavior within complex traffic environments is pivotal within Intelligent Transportation Systems (ITS). Though this technology plays a significant role in alleviating the prevalent operational difficulties in logistics and transportation systems, the precise prediction of vehicle trajectories still poses a substantial challenge. To address this, our study introduces the Spatio Temporal Attention-based methodology for Target Vehicle Trajectory Prediction (STATVTPred). This approach integrates Global Positioning System(GPS) localization technology to track target movement and dynamically predict the vehicle's future path using comprehensive spatio-temporal trajectory data. We map the vehicle trajectory onto a directed graph, after which spatial attributes are extracted via a Graph Attention Networks(GATs). The Transformer technology is employed to yield temporal features from the sequence. These elements are then amalgamated with local road network structure maps to filter and deliver a smooth trajectory sequence, resulting in precise vehicle trajectory prediction.This study validates our proposed STATVTPred method on T-Drive and Chengdu taxi-trajectory datasets. The experimental results demonstrate that STATVTPred achieves 6.38% and 10.55% higher Average Match Rate (AMR) than the Transformer model on the Beijing and Chengdu datasets, respectively. Compared to the LSTM Encoder-Decoder model, STATVTPred boosts AMR by 37.45% and 36.06% on the same datasets. This is expected to establish STATVTPred as a new approach for handling trajectory prediction of targets in logistics and transportation scenarios, thereby enhancing prediction accuracy.

Figures (6)

• Figure 1: Network structure and data handling processes. (a) is the structure of Transformer encoder, (b) is the structure of Transformer decoder, (c) is the network structure of this paper, (d) is the graph attention network structure, and (e) is the data processing process of filter layer.
• Figure 2: Diagram of the process of extracting roadway neighbor information.
• Figure 3: Map of the road network used in the experiment. (a) Beijing City Road Network and (b) Chendu City Road Network.
• Figure 4: The trajectory prediction results of STATVTPred. (a),(b) Predicted results for the Beijing dataset, (c),(d) Predicted results for the Chendu dataset.
• Figure 5: Comparison plot of the effect of different input feature dimensions on the experiment.