Solving Indoor Mapping and Navigation

点击翻译按钮获取中文摘要

Outdoor Mapping and GPS changed our lives. Hundreds of millions of people rely on the services everyday, driving to a cafe for a business meeting, planning a vacation trip in an unknown city, or taking a sick child to a hospital. We spend most of our time indoors. Indoor Mapping and GPS, if successful, would similarly make significant impact to our society in every aspect.This proposal seeks to establish robust computational frameworks for Indoor Mapping and GPS: 1) Producing structured geometric models of building interiors as floorplans that understand architectural components (e.g., walls and rooms) and their relationships (e.g., a door connects an office and a corridor); and 2) Tracking motions and localizing the position of an individual just from a smartphone anytime anywhere even in complete darkness. We seek to make breakthroughs by exploiting emerging mobile 3D sensing, low-energy inertial sensors, and big-data machine learning techniques. A key milestone will be an indoor GPS and Mapping system of Simon Fraser University campus, where one is able to localize oneself and find a direction to a destination indoors and outdoors, the first of a kind in the world at a campus scale.The potential impact of Indoor Mapping and GPS will be immense. People will never get lost indoors. Indoor navigation will enable the visually impaired to explore indoor spaces, and the mobility impaired to look up accessible routes. In the days of mobile computing, Indoor Mapping and GPS will open a completely new market for location aware services. Automatic building reconstruction will lead to effective asset creation for gaming, an active industry in many Canadian cities such as Vancouver, Toronto, or Montreal. Structured geometry reconstruction will lead to better map creation, critical for autonomous car industry emerging in Toronto. Potential applications of Indoor Mapping and GPS would go far beyond these examples, and include architecture, real estate, and construction.Outdoor GPS flourished via satellites in the 90s. Outdoor Mapping met a solution through digital imaging and 3D reconstruction techniques after the millennium. We believe that Indoor Mapping and GPS are about to find a way through emerging 3D sensing, wearable computing technologies, and deep neural networks, which will enable a plethora of novel applications in science, engineering, and commerce.