Vehicle Location Prediction System Development

Naidanov I. V. Vehicle Location Prediction System Development. 47 pages. 2021. This work is devoted to predicting the location of vehicles based on driver routes collected by city surveillance systems. The work contains 4 chapters, 8 tables, 25 illustrations. The first chapter describes two main mathematical models used for location prediction - markov chains and recurrent neural networks; considered models are compared and evaluated; the most accurate models for predicting the location of objects are determined. The second chapter describes the proposed methodology for predicting the location of vehicles based on driver routes collected by city surveillance systems. The third chapter describes the implementation of the proposed methodology on the data collected by Perm city video surveillance systems; several prediction models are developed. In the last chapter the web service for predicting location is developed; possible integration possibilites with "Safe city" information systems are considered. The results of this work could be useful to researchers who are considering to implement location prediction models for the "Safe city" information systems. Keywords: markov chains, recurrent neural networks, location prediction, city surveillance systems.