Demo: assisting visually impaired people navigate indoors

Abstract

Research in Artificial Intelligence, Robotics and Computer Vision has recently made great strides in improving indoor localization. Publicly available technology now allows for indoor localization with very small margins of error. In this demo, we show a system that uses state-of the-art technology to assist visually impaired people navigate indoors. Our system takes advantage of spatial representations from CAD files, or floor plan images, to extract valuable information that later can be used to improve navigation and human-computer interaction. Using depth information, our system is capable of detecting obstacles and guiding the user to avoid them.