Research on Short-Time Prediction of Dynamical Local Replanning Route Guidance Method Based on HMM

Abstract

The insufficient real-time responses, accuracy and intelligence have become key issues in the practical application of traffic guidance information services. This paper addresses these issues by proposing a new dynamic route guidance method. It firstly establishes a concurrent global route search method. By using this method, multiple relative static shortest routes can be searched, and then the shortest global optimized route is obtained for the current traffic flow. Secondly, by using the sliding window model, the method extracts the real-time traffic data stream reflected by the spatial and temporal changes in location of vehicles. By combining with the hidden Markov model, the method can also be used for the forecast of short-term traffic states and the decision-making of whether local planning is necessary.