PFI-TT: Crowdsourced Road Geometry Estimation using Smartphones

点击翻译按钮获取中文摘要

The broader impact/commercial potential of this Partnerships for Innovation - Technology Translation (PFI-TT) project is enabled by the development and prototyping of a novel smartphone-based, crowd-sensing system that can estimate road geometry features such as grade, curvature, elevation, cross slope, and superelevation. Such a system can potentially act as a cost-effective and scalable alternative to current road geometry acquisition systems which typically employ ground or aerial surveys using specially-instrumented vehicles/aircrafts. Information on road geometry can help in: a) improving driving-safety and efficiency for both human-driven and autonomous vehicles, b) improving smartphone/in-car localization and navigation services, especially in areas with limited global positioning system (GPS) accuracy, and c) conducting risk assessment and engineering design of road segments. The project seeks to address the unique research challenges introduced using smartphones as a sensing platform for the task of road geometry estimation. First, due to low quality sensors, data from smartphones is noisy and prone to drifts and biases. Second, to make the system transparent to users, the Smartphone must be able to be placed in any arbitrary position and orientation in the vehicle. Third, the crowd-sensing system is characterized by varying QoI (Quality of Information) of data from different sources/vehicles due to factors such as varying physical properties of vehicles, varying quality of sensors on different smartphones, etc. Successful implementation of the integrated solution will result in novel, heterogeneous sensor fusion algorithms to handle sensor noise and arbitrary placement by combining data from multiple types of sensors on the smartphone and data from auxiliary sources such as remotely sensed road elevation. The project will result in the development of data aggregation algorithms that combine multiple users’ smartphone data in a QoI-aware manner to address the unreliability of individual vehicle data.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.