Intelligent Service Robotics https://doi.org/10.1007/s11370-018-00272-5 ORIGINAL RESEARCH PAPER Development of an interactive game-based mirror image hand rehabilitation system Sangjoon J. Kim 1 · Sang Yun Han 2 · Gi-Hun Yang 3 · Jung Kim 1 · Bummo Ahn 3 Received: 29 May 2018 / Accepted: 27 December 2018 © Springer-Verlag GmbH Germany, part of Springer Nature 2019 Abstract To develop a hand rehabilitation device, the patient-specific heterogeneity of physical ailments must be considered in the design process to provide optimized rehabilitation. In this paper, we suggest a low-cost customized manufacturing process of a hand rehabilitation system. We first extract the length and size of the fingers based on a CMOS camera system. Then, the mechanical components of the rehabilitation system were manufactured using a 3D printer. User safety is guaranteed using a simple operation range control connector mechanism which mechanically locks up when the range of motion of each finger is exceeded for finger extension. We verified the usability of the hand rehabilitation system and applied the system to two custom interactive video games. Keywords Post-stroke hemiplegic patient · Hand rehabilitation · Soft glove · Mirror image exercise · Interactive game 1 Introduction Hand and finger paralysis, which may lead to the loss of the ability to partially or totally move, is caused by stroke, spinal cord injury, and cerebral palsy, which can greatly affect the activity of daily life of patients [1–5]. While optimal rehabil- itation method for hand and finger paralysis has not yet been clearly defined, the recovery is greatly affected by the sen- sorimotor experience [6] and repetitive task practice (RTP) from physical therapist [7–10]. However, RTP provided by physical therapists is labor intensive, time demanding, and costly. During the past decade, as a solution to ease such bur- dens, a great number of hand and finger robot rehabilitation systems have been introduced in the literature [11–19]. Chiri et al. [11, 12] developed a finger rehabilitation module using one actuator and three force sensors for each finger to exercise B Bummo Ahn bmahn@kitech.re.kr 1 Division of Mechanical Engineering, School of Mechanical, Aerospace and Systems Engineering, Korea Advanced Institute of Science and Technology, Daejeon, South Korea 2 Department of Industrial Design, Hanseo University, Seosan, South Korea 3 Robotics R&BD Group, Korea Institute of Industrial Technology (KITECH), 143 Hanggaul-ro, Sangnok-gu, Asan, South Korea the metacarpophalangeal (MCP) joint using a slider-crank- like mechanism. Wege et al. [13, 14] developed a hand exoskeleton that actuates each joint via a Bowden cable driven by an actuator, which is controlled by the electromyog- raphy (EMG) signals and used one motor for each joint. Ueki et al. [15] also presented a hand exoskeleton that is capable of 18 degree of freedom (DOF) motions. Jones et al. developed a 3 DOF (MCP, PIP (proximal interphalangeal), and DIP (dis- tal interphalangeal) joints) finger exoskeleton (AFX) for each finger using a cable mechanism [16]. Otto Bock Inc. [16] developed the WaveFlex that is a commercialized product capable of providing continuous passive movement (CPM) hand rehabilitation exercise. Takahashi et al. [17] developed the Hand Wrist Assistive Rehabilitation Device (HWARD) that can provide flexion and extension motion of the hand and wrist using 3 DOF pneumatic actuation system. Wu et al. [18] proposed a finger rehabilitation system using 2 DOF motions (flexion and extension) actuated by pneumatic system. Most of these systems were in the form of rigid robotic structures. Using rigid component, system complexity becomes a criti- cal issue as the hand and finger have high degrees of freedom. Consequently, most current forms of hand and finger reha- bilitation robots are leaning toward soft robotic technology [19–24]. Soft robotics uses soft and deformable materials to construct the mechanical systems instead of rigid frame or materials [25, 26]. The use of soft components removes con- straints on non-actuated degrees of freedom and also reduces 123