Map-Matching Techniques for Train Localization: A Taxonomic Survey

Precise localisation of rail vehicles is a key element towards the development and deployment of novel train control systems. Localisation of trains can be considered as a 1-D localisation problem, as trains only move on tracks, easing the localisation approach. Localisation with maps makes a great choice as the maps describe unambiguously train tracks, easing the train localisation and making it an effective choice. On top of that, train localisation with a map requires just onboard sensors making it a low-cost alternative to localisation with track-side equipment, as this equipment makes train localisation costs increase with the reach of the use. Even though this topic has been widely studied in the literature, there has not been a classification of the used methodologies for map-matching for train localisation. Therefore, the purpose of this paper is to make a classification of the state-of-the-art map-matching algorithms for train localisation. On top of that, this paper also discusses how to build a digital map from a set of coordinates. Three main categories have been identified in the literature for map-matching: geometric, similarity and hypothesis. And mainly three types of digital maps have been observed in the literature: interpolation, splines and geometric. This paper helps practitioners and researchers have a comprehensive foundation on map-matching for train localisation.