249 Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. SAICSIT '11, October 3–5, 2011, Cape Town, South Africa. Copyright © 2011 ACM 978-1-4503-0878-6/11/10... $10.00 The Impact of Sensor Fusion on Tilt Interaction in a Mobile Map-Based Application Bradley van Tonder Nelson Mandela Metropolitan University PO Box 77000, Port Elizabeth, 6031 South Africa +27 41 5042247 Bradley.vanTonder@nmmu.ac.za Janet Wesson Nelson Mandela Metropolitan University PO Box 77000, Port Elizabeth, 6031 South Africa +27 41 5042247 Janet.Wesson@nmmu.ac.za ABSTRACT Sensor-based interaction techniques relying on tilt gestures are increasingly being adopted in a wide range of mobile applications. Despite the recent increase in the availability of gyroscope sensors on mobile phones, most implementations of tilt interaction still rely on accelerometer input. Accelerometer sensors are susceptible to noise and also detect linear acceleration. This can lead to controllability problems, particularly when the user is mobile. This paper looks at how the fusion of accelerometer, gyroscope and digital compass sensor data impacts upon tilt interaction in a mobile map-based application. The results of an experiment comparing accelerometer-only tilt interaction with a sensor fusion approach are presented. The results show that while both the accelerometer and sensor fusion approaches performed well for seated tasks, the sensor fusion approach was shown to have advantages in terms of perceived controllability and efficiency while participants were walking. Categories and Subject Descriptors H.5.2 [Information Interfaces and Presentation]: User Interfaces – input devices and strategies. General Terms Design, Experimentation, Human Factors Keywords Tilt interaction, sensor fusion, mobile map-based applications, gyroscope, accelerometer 1. INTRODUCTION Mobile phones are increasingly being equipped with a wide array of sensors. Accelerometer, gyroscope and digital compass (magnetometer) sensors all provide important information regarding the orientation and movement of a mobile phone [13]. This information has allowed the development of interaction techniques which go beyond the traditional keypad and touch- screen interfaces. Tilt interaction allows users to interact with mobile applications by tilting the phone relative to the x and y axes. This interaction is used to perform panning and zooming in a wide range of applications. Despite the recent emergence of gyroscope sensors on mobile phones, most tilt interaction implementations rely on the use of accelerometers to determine the orientation of the mobile device [3, 12]. Accelerometers have several shortcomings when it comes to determining the orientation of mobile phones. Firstly, it is difficult to isolate acceleration as a result of tilting relative to gravity from linear acceleration as a result of movement [16]. Accelerometers are also susceptible to detecting acceleration as a result of hand tremor and other minor movements [2]. Gyroscopic sensors provide the potential to address many of the shortcomings of accelerometers, by measuring the angular velocity of the phone in three dimensions. Gyroscopic sensors, however, are susceptible to drift, which makes them unsuitable for stand-alone use in determining orientation. Sensor fusion, in which accelerometer, compass and gyroscope data is combined to determine device orientation, provides the potential to address the shortcomings of the individual sensors [4]. In this paper, the use of a sensor fusion approach, combining accelerometer, magnetometer and gyroscope data is investigated. The aim of this research is to determine whether a sensor fusion approach can improve the efficiency and controllability of an accelerometer-only tilt interaction implementation in the context of a mobile map-based application. The results of a user study are presented in which the two approaches were compared using a mobile map-based application in both seated and walking contexts of use. The results show that the sensor fusion approach provides several advantages in a walking context of use. 2. RELATED WORK Tilt interaction techniques allow users to interact with mobile applications by tilting the phone relative to the x and y axes. Tilt interaction has been used in a wide array of applications, from photo gallery applications to document browsers [2, 9]. Tilt interaction has recently started to emerge on mainstream mobile devices. Samsung and LG have both employed tilt interaction in applications where panning and zooming play an important role (such as photo and web browsing). Tilt angles have been used to calculate panning speeds, zoom levels and (in techniques such as Speed-Dependent Automatic Zooming [9]), both panning speed and zoom levels. Different approaches are used to map the tilt angles onto panning speeds. The two most commonly used approaches are rate control and