International Journal of Control, Automation, and Systems (2014) 12(6):1323-1335 DOI 10.1007/s12555-013-0241-3 ISSN:1598-6446 eISSN:2005-4092 http://www.springer.com/12555 Impedance Control of a Small Treadmill with Sonar Sensors for Automatic Speed Adaptation Jungwon Yoon*, Auralius Manurung, and Gap-Soon Kim Abstract: Automatic speed adaptation in treadmill training plays an important role in gait rehabilita- tion and virtual reality (VR) environments, where the user can adjust his/her speed for improved moti- vation and an enhanced sense of reality during walking interactions. To implement automatic speed adaptation of a treadmill belt, we have developed a novel impedance control scheme that accommo- dates natural movements without mechanical attachments to the user, and can estimate user-treadmill interactive forces to directly detect user intention, while simultaneously maintaining the user’s position on the treadmill platform. The proposed impedance control is realized via user interaction with a fixed virtual spring–damper component, allowing direct acceleration control of the treadmill belt in propor- tion to user displacement. The technique was applied to a small commercial treadmill (with a belt length of 1.2 m and a width of 0.5 m), which is easily installed and economical to operate, and is wide- ly used in homes and health centers. Inexpensive sonar sensors with a Kalman filter algorithm were employed to measure user motions. To identify the characteristics of the proposed control scheme, a set of experiments was conducted and preliminary user studies with VR interactions were performed. The results of these experiments indicate that our impedance control scheme can provide a non-intrusive, intuitive method for implementing user-selected speed on a small treadmill. The proposed technique is cost-effective, and could potentially be applied to any type of locomotion interface or gait rehabilita- tion system, without the use of expensive, sophisticated sensors or special treadmills. Keywords: Automatic speed adaptation, gait rehabilitation, impedance control, locomotion interface, treadmill. 1. INTRODUCTION A treadmill is commonly used in athletic training and clinical assessments such as gait rehabilitation. A tread- mill system that automatically adapts its speed to user intention can be very beneficial, since it allows users to interactively participate in training or simulations that involve walking. For virtual reality (VR) navigation, a treadmill is relatively safe, economical, and requires less space than other types of locomotion interface [1,2]. For rehabilitation purposes, treadmill training with a body weight support system has shown significant success in treating chronic non-ambulatory patients [3]. Recently, there has been increasing demand for self-selected speed adaptation with higher patient intention to enhance the effectiveness of gait training [4]. Most existing schemes for automatic speed adaptation of a treadmill can be categorized as user-position-based controls, user-force-based controls, or gait-parameter- based controls. One traditional method for speed adapta- tion of a locomotion interface [5] or gait rehabilitation system [6,7] is to update the treadmill speed via mea- surement of the physical interaction force between a user and a mechanical system. Since a user feels an inertial force proportional to his/her body weight during accele- ration on the overground, a control command for obtain- ing velocity updates can be derived from force measure- ments through a mechanical tether [5,6], or can be esti- mated by measuring the joint torques of a Lokomat [7]. This approach allows user intention to be directly ac- quired for speed updates, and can be effective for safe navigation of a locomotion interface and safe gait reha- bilitation training. However, it can also restrict the user’s natural motions during walking, and may not provide realistic proprioceptive feedback from mechanical cha- racteristics for normal users and highly ambulatory pa- tients [8]. It has recently been reported that measuring ground reaction forces on a separate treadmill can pro- vide a self-selected speed scheme [9] for a stroke patient with asymmetric gait, without severely restricting user motions. However, the technique requires an expensive instrumented treadmill and a complicated environment for user position measurements, which presents an ob- stacle to its widespread use. © ICROS, KIEE and Springer 2014 __________ Manuscript received May 14, 2013; revised February 8, 2014; accepted May 2, 2014. Recommended by Associate Editor Soohee Han under the direction of Editor Myotaeg Lim. This work was supported by the National Research Foundation Korea (NRF) funded by the Ministry of Education, Science and Technology (2012R1A2A2A01047344) and supported by Dual Use Technology Program of Civil and Military. Jungwon Yoon and Auralius Manurung are with the School of Mechanical and Aerospace Engineering and ReCAPT, Gyeong- sang National University, Jinju, Korea (e-mails: jwyoon@gnu.ac. kr, manurung.auralius@gmail.com). Gap-Soon Kim is with the Department of Control and Meas- urement Engineering, Gyeongsang National University, Jinju, Korea (e-mail: gskim@gnu.ac.kr). * Corresponding author.