An Approach to Geometry-Based Dynamic Location Referencing

An important requirement for knowledge infrastructures in smart cities is the continuous updating of location-based information. Protocols for dynamic location referencing like e.g. OpenLR or AGORA-C tackle the problem of accurately matching locations between dissimilar digital maps. They are map-agnostic and aim at limiting the amount of descriptive data to reduce bandwidth. However, there are applications for which the weaker requirement of map-independence is completely adequate, and for some there are even no restrictions in bandwidth (e.g. in the EC-funded project ROSATTE, and in the DLR projects MobiLind and KeepMoving), and with relaxed constraints it is possible to learn from methods in similar areas like road network matching and map conflation, in order to achieve a more accurate solution. Following this path, this paper presents a map-independent approach developed in the ongoing DLR project I.MoVe, which can be combined with a bandwidth-efficient dynamic location referencing method like e.g. OpenLR to target applications with bandwidth restrictions. It is currently capable of mapping closed line locations (i.e. circular routes, representing either the boundaries of areas or the tours of e.g. delivery trucks) from a TeleAtlas map to a NAVTEQ map on-the-fly with a success rate of 97.5% (OpenLR: only 82.5%), and also capable of mapping short line locations (i.e. linear routes) on-the-fly between the same maps, with a success rate of 99.7% (OpenLR: 91.9%)