Human navigation assistance with a RGB-D sensor

This paper focuses on the creation of a human navigation assistance prototype. The system uses a conventional RGB-D camera and a laptop to analyze the environment surrounding the user and provides him with enough information for a safe navigation. The system is designed to work indoors and performs two main tasks: floor and obstacle detection and staircase detection. Both tasks make use of the range and visual information captured by the sensor. The camera points downwards, allowing to acquire relevant navigation information without invading the privacy of other people. The system has been tested in real environments showing good results in the detection of obstacles and staircase.