A Cluster-based Framework for Predicting Large Scale Road-Network Constrained Trajectories

Abstract

The increasing availability of vehicle trajectory and road network datasets is crucial for the development of novel trajectory data mining-based applications. For instance, we can design more efficient routing protocols by applying vehicle trajectory prediction. In this paper, we propose a new cluster-based framework to predict road-network constrained trajectories. The framework, designed to perform long-term predictions, combines several steps that use historical trajectory datasets to train prediction models. Experimental results show the framework's effectiveness and efficiency to predict trajectories with different characteristics in a new real-world, large-scale scenario. Besides that, the framework outperformed some other solutions found in the literature in terms of prediction accuracy and computational overhead.mmm;