A Floor Plan based Vision Navigation System for Indoor Navigation with Smart Device

Many smart devices like smart phone and tablet nowadays are featured for hybrid sensor platform of GPS chip, inertial sensor(s), magnetic compass and other gadgets such as camera and Wi-Fi. The interest to apply those smart devices for indoor navigation is growing since a large variety of sensors on such devices enable hybrid location solutions to not only improve the availability of indoor positioning but also the accuracy and smoothness. However, in deep indoor scenario, the positioning accuracy is still seldom satisfactory due to large accumulative errors of dead-reckoning sensors. In this paper, a floor plan based vision navigation method is designed for pedestrian handset indoor application. The floor plan for buildings is an easily accessible indoor map with detailed path and room information. It can be matched with the vision measurements from the camera sensor to derive accurate and drift-free positions even in deep indoor environments. The Random Sample Consensus (RANSAC) algorithm is adopted for robust matching between floor plan and camera photo. An iPhone Demo App is developed to evaluate the performance of the designed system and the test results indicate meter-level horizontal accuracy.