Route Matching Research Based on Roadless Navigation Data Improvements on Hidden Markov Model

Abstract

The positioning problem is a key pre-processing step in the location service to assist in solving related problems such as location query and sharing in urban problems. Most of the map matching results obtained in these research fields are based on road networks. However, it is not cost-effective to buy a set of road network data in a country with a small amount of business, and in areas where the road network is thinly scattered and the maintenance work of road network is not timely enough. At this time, it is a big challenge to position the vehicle on the predetermined route basing on roadless navigation data. Inspired by road network-based map-matching algorithms, the contribution of this paper is introducing the time constraints and the direction constraints to transition probability in Hidden Markov Model(HMM), so that the position of the vehicle can be more accurately located at a certain point in the route, and the application of the algorithm in this no road networks scenario is improved accordingly. Experiments show that, in the case of the high sampling rate, the accurate performance meets the business requirements.