III: Small: Towards Highly Accurate Map Services

点击翻译按钮获取中文摘要

Map services, including finding the route to a certain destination or searching for nearby facilities, have empowered a variety of ubiquitous services and applications, including transportation, location-based services, ride-sharing, and a broader delivery industry. A major assumption for all such services is that the underlying map is accurate, which is not always true. Maps suffer from all sorts of inaccuracies that impact the optimality of map services. While this may be acceptable for causal users where inaccuracy may only mean few minutes of delay, it is not the case for applications that support map services for large numbers of users. Recent reports show that map inaccuracy in the USA costs delivery companies alone billions of dollars annually. This trend is just going to increase with the increase of online shoppers and riders, which shifts traffic from casual users to delivery companies and ride sharing services. This project aims to conduct research, develop requisite knowledge, and build software infrastructure that will significantly boost the accuracy of the underlying map, which will have a direct impact on map services efficiency.The project aims to achieve its goals through exploiting two orthogonal directions. First, building the most comprehensive and accurate maps worldwide, with a broad definition of a map that goes beyond the traditional map topology to a richer definition that includes traffic-related metadata, without which, most of map services will have low accuracy. This is achieved through harnessing trajectory data and open-source maps to identify and infer missing information. Second, developing new techniques that would increase the accuracy of query processors over the accurately built maps. This is achieved through understanding the error patterns of query processors and using it to calibrate the query results, while taking advantage of the accurately built maps and their metadata.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.