Matching Traffic Lights to Routes for Real-World Deployments of Mobile GLOSA Apps

Green Light Optimized Speed Advisory (GLOSA) apps provide speed recommendations for drivers to pass traffic lights during their green phases. In this way, the comfort and efficiency of traveling can be significantly improved. Thus, GLOSA apps are a valuable contribution to smart mobility. Mobile GLOSA apps provide an attractive alternative to static info signs, but they need to anticipate upcoming traffic lights that the vehicle will pass. While this imposes no challenge for predominating research within simulation or test track environments, real-world deployments need to correctly match a few from thousands of traffic lights to a route. In this paper, we discuss in a novel approach that MAP topologies, an international ETSI standard for turn geometries of traffic lights, can be used to perform this matching. However, routing is usually performed on public map data, which is not aligned with the MAP topologies. We explore two computational methods, specifically map-matching as preprocessing for adjacency lookup and topologic feature matching, that account for discrepancies between the MAP topologies and the route. We show that the core problem can be addressed using these algorithms to enable large-area deployments of real-world mobile GLOSA apps. In a comparative evaluation, the topologic feature matching technique achieved an F1 score of 89.5%, while the map-matched adjacency lookup method only achieved an F1 score of 48.3%. We analyze this performance gap and conclude further research directions.