Technology for Arm Training

We have developed several devices for arm movement training after stroke and used these devices to study how people with stroke can best retrain their arm movement. This page describes the evolution of this work, from early testing with a trombone-like device called the ARM Guide, to a low-cost internet-based system called Java Therapy, to a passive exoskeleton called T-WREX which is now sold by Hocoma A.G. as ArmeoSpring, to sophisticated robotic arm exoskeletons Pneu-WREX and BONES.

pneu Wrex

ARM Guide

One of the first devices we developed was the ARM Guide, a trombone-like device that can assist in arm movement in different directions. This project was supported by NIDRR and was collaborative with Lennie Kahn and Zev Rymer at the Rehabilitation Institute of Chicago. We used this device to test the hypothesis that active-assisted arm exercise after chronic stroke would be more effective than unassisted exercise. Our first study with twenty stroke subjects at the Rehabilitation Institute of Chicago suggested that actively assisting in movement in fact did not improve movement recovery more than a matched amount of free reaching exercise (see Kahn et al. 2006). This suggests that repetitive movement attempts by the patient is a primary stimulus to movement recovery (rather than motorized assistance from the robot). In fact, we have later found that assisting people with a robot can cause people to slack, or in other words decrease their motor output, which may decrease motor learning (see Wolbrecht et al. 2008 and Emken et al. 2007.

Java Therapy

With support from Microsoft, we also previously developed a low-cost approach to home-based therapy called Java therapy. Java Therapy was a web site that allowed people with a stroke to practice simple movement training exercises using a low-cost force-feedback joystick at home, and to receive quantitative feedback about their movement recovery progress. People with stroke found the system engaging, but improvements in movement ability seemed limited to the movements practiced, which were joystick movements rather than more functional activities of daily living. See Nessler et al. 2002


To allow more functional arm training we developed a device called T-WREX (or “Therapy-Wilmington Robotic Exoskeleton”), which was the doctoral dissertation research of Dr. Robert Sanchez and was work supported by the National Institute of Disability and Rehabilitation Research (NIDRR). We designed T-WREX to be non-robotic, but to still allow severely weakened patients to move by providing gradable assistance against gravity with elastic bands. To achieve this, we collaborated with Dr. Tariq Rahman of the A.I. DuPont Institute for Children, who with NIDRR support had developed the innovative arm support called WREX to assist children with weakened arms in moving their arms. We scaled the WREX design to be large enough and strong enough to support movements by adults with a stroke. We also designed T-WREX to support functional movements. The use of WREX helped achieve this goal in part because WREX allowed a large range of motion and had been explicitly designed to allow feeding and other functional movements. But we also developed and integrated a grip sensor that allowed detection of even trace amounts of hand grasp, thus allowing people with weakened, essentially “useless” hands to practice using their hands in a meaningful way in a virtual world, in coordination with their arms. We also developed new computer games that were easy to learn yet engaging and which approximated the movements needed for activities of daily living. These games included activities such as cooking, shopping, bathing, and cleaning.

Through a clinical test of 28 chronic patients at the Rehabilitation Institute of Chicago, run by Occupational Therapist Sarah Housman, we found this approach to be motivating, engaging, and therapeutically effective (see Housman et al. 2009). Read more on the MARS Rehabilitation Engineering Research Center (RERC) web site. A modified version T-WREX was commercialized by Hocoma as ArmeoSpring, and is a successful product now used by almost 200 sites as of 2011. The first research study performed with the device, outside of our own work, was published in 2011. This study found that training with ArmeoSpring improved arm function of people with Multiple Sclerosis. (see: Gijbels et al. 2011)

The role of physical assistance in helping people recover arm movement ability is unresolved. Assistance helps motivate people to practice moving, and provides a richer sensory experience, but some forms of assistance apparently have a negative effect because they cause slacking and reduce movement variability. The use of a gravity balance approach to provide assistance, as used with T-WREX and ArmeoSpring, seems to strike a good compromise of improving self-efficacy and sensory input by allowing greater active range of motion for weakened patients, while limiting slacking and still allowing trajectory variability.


Working with robotics and control expert Prof. Jim Bobrow and neurologist Prof. Steve Cramer at U.C. Irvine, we are currently developing and testing sophisticated robotic exoskeletons for arm movement training in virtual environments with support from NIBIB and NCMRR at NIH. Movies of the robots are available here.