Potential of crowdsourced data for integrating landmarks and routes for rescue in mountain areas

ABSTRACT

Many different websites offer the opportunity to share and download landmarks and routes produced by the crowd. Landmarks near to a route or routes passing near to some landmarks may help in the context of mountain rescue. Therefore, it is necessary to identify relevant data sources and to describe their characteristics. In this paper, we set out to explore the potential of crowdsourced data in order to be considered such as data sources in the context of mountain rescue. Thus, our aim is to study the content of different sources to have a better knowledge on how landmarks and routes are mapped, to demonstrate the complementarity of crowdsourced data with respect with authoritative data, and to study the feasibility of defining links between routes and landmarks. The proposed method used integration techniques such as map matching, route construction and data matching. Among the results, the large number of non-matched features proves the richness of crowdsourced data. The matching results generate new semantic rules for both type of landmarks and geometries of route.

RÉSUMÉ

De plus en plus de sites web proposent de partager et de télécharger, entre amateurs de sport de plein air, des traces GPX d’itinéraires ainsi que des points de passage pouvant être utilisés comme points de repères. La connaissance des points de repères le long des itinéraires ou des itinéraires qui passent près de points de repères peut aider les secouristes dans la phase de localisation d’une victime en montagne. Afin de constituer des données métier (i.e. secours en montagne), il est nécessaire d’étudier la pertinence des différentes sources de données et de définir leurs caractéristiques. Dans ce papier, nous explorons le potentiel des données issues du web afin de déterminer leurs bien-fondés dans le contexte du sauvetage en montagne. Notre objectif est donc d’étudier le contenu de différentes sources de données afin de mieux comprendre comment les points de repères et les itinéraires sont saisis, de démontrer la complémentarité des sources de données en comparaison avec des données d’autorité et d’étudier la possibilité de définir des liens entre les itinéraires d’une part et les points de repères d’autre part. La méthode proposée utilise les techniques d'intégration telles que le recalage de points GPS sur un réseau routier, la reconstruction d’itinéraires et d'appariement de données géographiques. Un premier résultat concerne le grand nombre de données non appariées, ce qui prouve une grande richesse et complémentarité des sources. De plus les résultats de l’appariement de données font émerger de nouvelles règles sémantiques concernant les points de repère et les géométries des routes.

KEYWORDS:

• Crowdsourced landmarks
• crowdsourced routes
• data and map matching
• mountain rescue
• data integration

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Marie-Dominique Van Damme is a research engineer at the LASTIG laboratory of IGN, the French Mapping Agency. She undertakes studies and research in the field of spatial data integration, spatial data mediation, and spatial data quality. She is in charge for implementing prototypes (data matching web application, crowdsourcing software to collect data in the field) and tools for research purposes. She coordinates development projects for students from ENSG School.

Dr. Ana-Maria Olteanu-Raimond is researcher at the LASTIG laboratory of IGN, the French Mapping Agency. She is interesting in spatial data integration, spatial data quality assessment, uncertainty modeling, data matching, land cover/land use update, crowdsourcing and citizen science. She is the coordinator of the French project CHOUCAS project (choucas.ign.fr) aiming at proposing methods and tools to localize victims in mountain area.

Yann Méneroux is a PhD student at the LASTIG laboratory of IGN, the French Mapping Agency. His research interests are currently focusing on machine learning techniques for map inference of road signs based on probe vehicle data and GPS trajectory preprocessing algorithms.

Notes

1. GR is a well know European abbreviation defining a long distance path used for hiking activities. GR20 is a famous GR path in Corsica having 179 km length.

2. https://github.com/ANRChoucas/PluginChoucas

3. https://github.com/IGNF/MultiCriteriaMatching

4. https://github.com/IGNF/mapmatcher

Additional information

Funding

This work was supported by Agence Nationale de la Recherche [grant number ANR-16-CE23-0018].