Effect of Camera Characteristics on the Accuracy of a Visual Gyroscope for Indoor Pedestrian Navigation Laura Ruotsalainen*, Jared Bancroft and Gérard Lachapelle *visiting from Finnish Geodetic Institute PLAN Group, Department of Geomatics Engineering University of Calgary Calgary, Canada Heidi Kuusniemi and Ruizhi Chen Department of Navigation and Positioning Finnish Geodetic Institute Kirkkonummi, Finland Abstract— Accurate navigation in GNSS degraded environments, namely indoors and in urban canyons, is still an unsolved problem. Other radio positioning systems used for indoor navigation purposes, like WLAN, are dependent on an infrastructure. Alternatively, if the initial position of the user is known, it may be propagated using self-contained sensors, such as digital compasses, rate gyroscopes and accelerometers. However, rate gyroscopes used for measuring angular velocity suffer from drift errors, and need augmentation to provide accurate positioning. Visual aiding is a substantial method for augmenting angular velocity measurements because it suffers from errors of a different nature than those of rate gyroscopes. The concept of a “visual gyroscope” described in this paper is based on tracking the motion of vanishing points in consecutive images and translating the motion information into the rotation of the camera. The motion of the camera may be further transformed into the heading change of the user. This paper assesses different cameras and setup characteristics defining the accuracy and functioning of a visual gyroscope using three different cameras and two setups. Keywords: visual gyroscope, vanishing point, indoor pedestrian navigation, heading change I. INTRODUCTION Accurate positioning in indoor environments, even when Global Navigation Satellite Systems (GNSS) signals are degraded, is critical for first responders and desirable for personal navigation applications. Other radio navigation solutions, such as Wireless Local Area Networks (WLAN) or Bluetooth, provide absolute positioning in GNSS degraded environments, but are dependent of an appropriate infrastructure that is not always available and, if available, may not be functioning in case of fire and power outages. Self- contained sensors are commonly used to measure the relative position of the user [1]. When the user speed provided by accelerometers and angular velocity provided by rate gyroscopes are integrated with an initial location and heading, the user position may be propagated for a short period of time. The drawback of rate gyroscopes is drift errors that reduce the attitude accuracy substantially. The concept of a “visual gyroscope” has been introduced in e.g. [2], [3] and [4]. Visual gyroscopes are methods based on the use of computer vision algorithms to transform information found from images into the camera rotation. The rotation of the camera may be further advanced into the heading change of the user when the alignment of the camera to the user body is known. The visual gyroscope used in this paper is based on tracking vanishing points in consecutive images [4]. Vanishing points are points in an image where the lines parallel in the real world seem to intersect. When the camera is calibrated, motion of the vanishing points may be converted into rotation. Human- made constructions, like indoors and urban areas, are mainly composed of orthogonal structures. These structures form scenes consisting of straight lines parallel in three orthogonal directions. When an image is taken in such a scene, the lines in corresponding orthogonal directions seem to have three intersection points (vanishing points). Two of these vanishing points, namely the central and vertical vanishing points, are used to resolve roll, pitch and heading change. The visual gyroscope may further be used as an additional method to provide the user attitude and therefore mitigate the gyroscope errors deteriorating the performance of the navigation system. The results obtained by integrating the visual gyroscope with an Inertial Measurement Unit (IMU) show significant improvements on the user attitude and position outputs [5]. Image quality, camera rate and mounting location affect the performance of the visual gyroscope. The visual gyroscope does not suffer from drift errors, but its performance is reduced in the situations of low light, scenes lacking straight lines and sharp turns. This paper evaluates the characteristics of three cameras used as visual gyroscopes and their effect on accuracy in indoor pedestrian navigation. A very challenging environment was selected as a test area to assure the usability of the method. The three cameras used for the experiments are a GoPro HD Hero helmet camera [6] directed for first responders, a Sony HD video camera aimed for extreme sports [7] and a Nokia N8 smartphone’s camera [8]. Experiments consider the effects of different image sensors and different image rates on the accuracy of the visual gyroscope. The effect of free camera rotation is also assessed for the smartphone’s camera. The experiments consist of two different setups on equal periods of 30 minutes, each in within buildings at the University of Calgary campus. The results from experiments show that the image rate and characteristics of the lenses affect the accuracy of the visual gyroscope, namely a wide-angle lens with a small f-number produces images suitable for the method. The GoPro and the Nokia N8 cameras surpass the Sony camera in terms of success of calculations due to their 1 2nd International Conference on Ubiquitous Positioning, Indoor Navigation and Location-Based Service, Helsinki, 2-5 Oct 2012 ©2012 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE