A Real-time Indoor Visual Localization and Navigation Method Based on Tango Smartphone

Visual localization and navigation is a renew research hotspot in the field of indoor location based service (LBS). In order to address the problem of the real scale of indoor space cannot be determined based on the monocular vision based on smartphone, and the feature-based method performs poor in environments with weak texture. In this paper, a fusion of feature-based method and direct method based on Tango smartphone is proposed. It uses the direct depth information from Tango sensor to determine the real scale of indoor space, and combines the direct method in regions with little texture feature to estimate the motion of Tango camera. With the ability of distance between the Tango camera and the object can be obtained by Tango smartphone directly, to improve the accuracy and success rate of camera pose tracking through combined with feature-based method and direct method for visual localization and navigation in a complex space environment. Experimental results demonstrate that proposed method is more reliable and robust, and can achieve an stable real-time camera pose estimate in feature deficiency regions, where the relative translation error is in centimeter level, which can meet the needs of real-time indoor visual localization and navigation based on Tango smartphone.