Research on high-fidelity reconstruction algorithm for vehicle trajectory based on centralized spatiotemporal feature fusion

Abstract

Vehicle trajectory data, primarily characterized by time-series features, is the key information carrier for vehicle movement. It documents dynamic characteristics such as position, velocity, and acceleration across spatiotemporal dimensions. This data supports critical tasks, including microscopic traffic flow modeling, driving state analysis, and traffic safety assessment. To address two key challenges — low-sampling-rate trajectory data losing high-frequency motion details in intelligent transportation systems, and traditional methods failing to simultaneously meet requirements for high precision and motion plausibility — this study proposes a high-fidelity vehicle trajectory reconstruction algorithm based on centralized spatiotemporal feature fusion. Based on the SOFTS (Series-cOre Fused Time Series) framework, the algorithm constructs a global-local collaborative spatiotemporal feature representation mechanism via a multidimensional enhanced STAR (Spatio-Temporal Aggregation and Representation) module. Specifically, it incorporates a spatiotemporal embedding layer to capture interdependencies between timestamps and spatial coordinates. Residual connections preserve original trajectory details, while a spatial proximity weighting mechanism optimizes core feature aggregation. A dynamic weight matrix is constructed to adaptively focus on velocity-position correlations among neighboring vehicles within a 20-meter radius. Kinematic constraints are integrated to ensure the physical plausibility of reconstructed trajectories, including a kinematic loss function. This function uses acceleration smoothness regularization terms and trajectory curvature continuity regularization to guide the model toward physically feasible solutions that comply with vehicle dynamics. To validate effectiveness, the proposed method was extensively tested on public datasets (NGSIM, HighD, and CQSkyeyeX) and systematically compared with traditional approaches (e.g., linear interpolation) and deep learning models. Experimental results demonstrate that the improved algorithm significantly outperforms baseline methods in interpolation accuracy, spatiotemporal smoothness, and computational efficiency. The findings can be applied to high-frequency trajectory generation for autonomous driving, microscopic traffic flow simulation, and other domains, providing critical technical support for upgrading intelligent transportation systems from data collection to decision-making optimization.