CollabMap: Augmenting Maps Using the Wisdom of Crowds

Introduction The creation of high fidelity scenarios for disaster simulation is a major challenge for a number of reasons. First, the maps supplied by existing map providers (e.g., Ordnance Survey,1 TeleAtlas) tend to provide only road or building shapes and do not accurately model open spaces which people use to evacuate buildings, homes, or industrial facilities (e.g. the space around a stadium or a commercial centre both constitute evacuation routes of different shapes and sizes). Secondly, even if some of the data about evacuation routes is available, the real-world connection points between these spaces and roads and buildings is usually not well defined unless data from buildings’ owners can be obtained (e.g. building entrances, borders, and fences). Finally, in order to augment current maps with accurate spatial data, it would require either a good set of training data (which is not available to our knowledge) for a computer vision algorithm to define evacuation routes using pictures (working on aerial maps) or a significant amount of manpower to directly survey a vast area. Against this background, we develop a novel model of geospatial data creation, called CollabMap, that relies on human computation. CollabMap is a crowdsourcing tool to get users contracted via Amazon Mechanical Turk or a similar service to perform micro-tasks that involve augmenting existing maps (e.g. Google Maps or Ordnance Survey) by drawing evacuation routes, using satellite imagery from Google Maps and panoramic views from Google StreetView. In a similar vein to (Von Ahn, Liu, and Blum 2006; Heipke 2010), we use human computation to complete tasks that are hard for a computer vision algorithm to perform or to generate training data that could be used by a computer vision algorithm to automatically define evacuation routes. In so doing, we advance the state of the art in the following ways. First, we propose the first crowdsourced mapping system that relies on large numbers of non-experts to generate maps that define and connect open spaces (occupied by pedestrians or vehicles) to buildings and roads.Second, we extend the Find-Fix-Verify pattern by Bernstein et al. (2010) to include measures of trust and reputation of the task performers. Third, we propose a number of evaluation mechanisms for