Reconstruction Method for Multi-Vehicle Trajectories on Arterials Driven by Multi-Source Data

Vehicle trajectories contain enriched spatial and temporal traffic information. In this study, the vehicle trajectory data were obtained after the data-fusion process of multi-source heterogeneous data on arterials. Both the piecewise cubic Hermite interpolation algorithm and cubic spline interpolation algorithm were used to reconstruct the single vehicle trajectories. A cross-validation method was applied in the comparison for obtaining the optimal model. Based on the reconstructed vehicle trajectories, an interpolation method was used to predict the unrecorded multi-vehicle trajectories by interpolating the time of unknown vehicles. The results show that the piecewise cubic Hermite interpolation can achieve better performance in reconstructing the single-vehicle trajectory and it is effective in predicting the missing trajectories. This study supports the spatial-temporal analysis of vehicle trajectories, traffic-state estimation, and transportation optimization.