Evaluation of an Inexpensive Depth Camera for Passive In-Home Fall Risk Assessment Erik E. Stone Department of Electrical and Computer Engineering University of Missouri Columbia, MO, USA ees6c6@mizzou.edu Marjorie Skubic Department of Electrical and Computer Engineering University of Missouri Columbia, MO, USA skubicm@missouri.edu Abstract— We present an investigation of a new, inexpensive depth camera device, the Microsoft Kinect, for passive fall risk assessment in home environments. In order to allow older adults to safely continue living in independent settings as they age, the ability to assess their risk of falling, along with detecting the early onset of illness and functional decline, is essential. Daily measurements of temporal and spatial gait parameters would greatly facilitate such an assessment. Ideally, these measurements would be obtained passively, in normal daily activity, without the need for wearable devices or expensive equipment. In this work, we evaluate the use of the inexpensive Microsoft Kinect for obtaining measurements of temporal and spatial gait parameters as compared to an existing web-camera based system, along with a Vicon motion capture system for ground truth. We describe our techniques for extracting gait parameters from the Kinect data, as well as the advantages of the Kinect over the web-camera based system for passive, in-home fall risk assessment. I. INTRODUCTION o allow older adults to continue living longer in independent settings, and thus reduce the need for expensive care facilities, low-cost systems are needed to detect not only adverse events such as falls, but to assess the risk of such events, in addition to the early onset of illness and functional decline. Continuous, ongoing assessments of physical function would help older adults live more safely in independent settings, while also facilitating targeted medical interventions when needed. Ideally, such measurements would be obtained passively, in the course of normal daily activity [1]. This work focuses on developing a robust, low-cost, vision based monitoring system for assessing fall risk, detecting falls, and detecting the early onset of illness and functional decline. Research has shown the importance of measuring a person’s gait, including identifying stride-to- stride variability as a predictor of falls [2-4]. Vision based monitoring systems have the resolution needed to yield the detailed measurements of physical function necessary for fall risk assessment (and early illness detection) passively, in the home environment, on a continuous basis. Furthermore, research has shown that the privacy concerns of older adults to video based monitoring systems can be alleviated through appropriate handling and processing of the video data, e.g., in the form of silhouettes [5]. Recently, Microsoft has released a new, inexpensive device, called the Kinect, to allow controller free game play on their Xbox system. The device uses a pattern of actively emitted infrared light to produce a depth image (the value of each pixel depends on the distance of what is being viewed from the device) which is invariant to visible lighting; and, thus, allows for a 3D representation using a single Kinect. This technology offers a number of potential benefits for low-cost, vision based monitoring systems. This paper presents an investigation of the Kinect as a sensor for fall risk assessment. Specifically, techniques for acquiring spatial and temporal gait parameters from the depth data of the Kinect are presented; along with a comparison between the measurements obtained from the Kinect, to those obtained from an existing web-camera based system, and a Vicon marker based motion capture system. II. BACKGROUND Recent research in activity monitoring of older adults has focused on the use of passive infrared (PIR) motion sensor suites in the home [6-7]. These sensor suites yield information about the daily activity levels of monitored subjects, and arrays of such sensors have been used to obtain velocity measurements on a continuous basis in home settings [8]. While such systems don’t raise privacy concerns among older adults, they typically do not produce measurements of the detail necessary for assessment of fall risk; specifically, spatial and temporal gait parameters (other than walking speed), timed up and go (TUG) time, sit to stand time, etc [9]. Existing systems for capturing such measurements are typically wearable, accelerometer based devices, expensive gait or motion capture systems, or direct assessment by a health care professional [2]. Wearable accelerometer based devices for obtaining detailed measurements of physical activity, specifically gait parameters, is an area that has been the focus of much research [10]. Efforts have even included utilizing accelerometers in existing smart devices, which individual’s may already own and potentially carry with them. However, many elderly adults are reluctant to use wearable devices because they consider them to be invasive or inconvenient [1]. Thus, wearable devices may not be reliable for capturing movement in the home for continuous monitoring and T PervasiveHealth 2011, May 23-26, Dublin, Republic of Ireland Copyright © 2012 ICST DOI 10.4108/icst.pervasivehealth.2011.246034