            JOSIP MUSIĆ  , MOJMIL CECIĆ  , MIRJANA BONKOVIĆ   Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture University of Split R. Boškovića b.b. CROATIA jmusic@fesb.hr, mcecic@fesb.hr, mirjana@fesb.hr http://labacs.fesb.hr   In the paper we propose a novel hands3free pointing device for the disabled based on commercially available inertial sensors (XSens MTx) and evaluate its performance on twelve test subjects using multi3directional point3and3select task and experimental design adopted from [1]. In experiments throughput was chosen as a main performance parameter, with error rate being the additional parameter. Two pointing modes were tested, as well as two selection techniques. Additionally, all test subjects were given questionnaires in which they had to rate pointing device comfort and performance. Obtained results are presented and discussed.   inertial sensors, head3joystick, throughput, performance evaluation, XSens   Personal computers are ubiquitous in everyday life, and with development and application of more and more sophisticated programs and with new and innovative applications, have the ability to increase life quality and work efficiency of the individual. An important parameter which needs to be taken into account when considering new applications of personal computers is targeted user group and related constraints. For example, dentists and surgeons have the need to use computers while their hands are occupied which is impossible with computer mouse or keyboard. Also, people with certain types of disabilities (e.g. quadriplegia) are unable to use computers and are thus denied numerous benefits both in their rehabilitation and everyday life. Our research is focused on input devices for the people with disabilities and our ultimate goal is to design and develop inertial sensor based computer input system which we named HeadJoystick and which would be used to control computer and other devices (e.g. wheelchair, robotic manipulator). In the literature number of research articles can be found on the topic of hands3free computer interface for the disabled. In research head pose is often used as a control parameter with head angles (roll, pitch, yaw) being transformed into screen coordinates. Manner in which head angles are measured differs from one article to the other and includes video based techniques [2], tilt sensors [3] and ultrasonic sensors [4]. Another approach utilizes voluntary muscle control of the disable person (usually face muscles) to control the pointer [5]. Activation of certain muscle or sequence of muscles triggers predefined control action(s). Certain system [6, 7] use magnetic sensors usually placed inside subject’s mouth to detect changes in magnetic field caused by movement of small permanent magnet attached to the tip of subject’s tongue. In this manner relative tongue position inside the mouth can be determined and appropriate control action taken. Eye tracking systems are also very popular [8]. They usually implement infrared cameras (although other approaches exist such as measurement of electrooculographic potential) to detect and track pupil movements and using data processing algorithms determine eye gaze position. Once eye gaze position is determined pointer is moved to that location. Systems based on measurement of EEG signals can also be used as human3computer interfaces for the disabled [9]. These systems sometimes called brain3computer interfaces (BCI) measure brain activity (brain waves) using either surface electrodes (non3 invasive) or surgically implanted electrodes (invasive). Measured brain signals are then mapped into control space and corresponding control action executed. The decision which system to use for particular individual is complex [10] and is based on number of variables some of which are: level of user disability, user preferences, user friendliness, learning curve and price. We believe our system to be user friendly, easy to use and is suitable for number of medical conditions and application scenarios (it doesn’t require any outside signal sources). The main disadvantage is sensor price (>1000 USD). The article is structured as follows. In section two experimental setup and procedures are described, and evaluation parameters defined. Section three presents and analyses obtained experimental results. Finally, some conclusions are drawn based on experimental Proceedings of the 10th WSEAS International Conference on AUTOMATION & INFORMATION ISSN: 1790-5117 163 ISBN: 978-960-474-064-2