Creating a Unique Rehabilitation Robot

A new approach to an old challenge

May 18, 2006

8 Min Read
Creating a Unique Rehabilitation Robot

Originally Published MPMN May 2006


Creating a Unique Rehabilitation Robot

A new approach to an old challenge

Joyce Laird

The ARC3D system consists of the modified Haptic Master with Biodex chair. The subject’s arm weight is supported to allow free movement.

Taking products that are available on the market and using them to create something new and better is the foundation of the best in medical device design technology. That is what happened when Jules Dewald, PhD, associate professor of Physical Therapy & Human Movement Sciences, Physical Medicine, and Rehabilitation and Biomedical Engineering at Northwestern University (Chicago; www. sulu.smpp.northwestern. edu), and Wim Lam, the owner and general manager of Lam Design Management (Orchard Park, NY), joined forces to develop a new type of robotic system for use in the rehabilitation of stroke victims’ limbs.

“Wim has the engineering and I have the scientific background,” says Dewald. “It was an ideal match. Together, we developed this new system and a new business, D.L. Rehab Technology.”

Creating the system was a blended process. Dewald had been doing in-depth research in stroke rehabilitation for years. He had conducted research using load cell sensors developed specifically for his needs by JR3 Inc. (Woodland, CA; to measure isometric elbow and shoulder torques of the affected upper limbs of stroke survivors. He wanted to take it one step further and use the same load cells to monitor subjects when actually reaching out while lifting the weight of their affected arms. This is something most stroke subjects find very difficult to do unless the arm is fully supported, such as by using an air-bearing device that slides over a large table.

Lam’s company represented a product from the Netherlands, a robot called the HapticMaster. When Lam demonstrated the system to Dewald, the two started throwing ideas around. Lam suggested that perhaps if the robot was used it could replace a real table with a virtual one.

Understanding the Challenge of Stroke Victims

“What we wanted to do was create a virtual world for stroke victims where the weight of their arm could be eliminated and then gradually reintroduced,” Dewald says. “A robot can eliminate the weight of the upper limb as the subject reaches out.”

Healthy people can lift up their arms against gravity and then move their joints—elbow, wrist, and fingers—independently while holding their arms away from their bodies. This allows actions such as reaching to a shelf, grasping an object, and then moving it to another spot.

“Even when weight is applied to a normal arm, any person can do this without a problem,” Dewald explained. “However, stroke victims can no longer drive one joint at a time. If they lift an arm they also flex the elbow, wrist, and fingers and their hand will pull into their chest. They have a very hard time extending their elbows. If you take the lifting out of the exercise, stroke victims regain control of their elbows. This permits stroke patients to extend their arms in the workspace as if reaching for an object.”

Putting the Pieces Together

The JR3 load cell was the key to the end effector in the ACT3D.

System development began in the summer of 2003. Using the best of two existing pieces of rehabilitation equipment with a new interface developed around the JR3 load cell, the first Arm Coordinated Training Device in Three Dimensions (ACT3D) began to take form.

“We compared Jules’s requirements with the capabilities of the standard HapticMaster robotic system and found that it would not suit what he was looking for, particularly not the end effector,” Lam says. “We not only needed to measure the three forces in the three linear axes, but also the torques. It was necessary to create a way to interface with the load cell and specially design an end effector to meet these requirements.”

The engineering design and building of these changes were done by Lam and Moog-FCS Robotics, manufacturer of the HapticMaster. The JR3 load cell was the key to the end effector. Based on the data from experiments, the sensor selected was a 4.5 × 1.5-in. 100-lb load cell which was integrated into the end effector.

The second main part of the system was a chair created by Biodex Medical Devices (Shirley, NY; www. The chair belongs to the company's single-link robotic system, and is often used for sports rehabilitation, where only one axis of rotation is required, as when athletes are strengthening their quadriceps following knee surgery. In the ACT3D, the chair is ideal because it provides a high level of control and repeatability for subject orientation with respect to the HapticMaster.

“It's a beautiful chair and incorporates an adjustable seat that can move on a track that can be adjusted to accommodate different-sized individuals,” says Lam. “This allows them to be placed in the correct position with respect to the robot. Instead of using the Biodex robot on it, we've connected it to our modified HapticMaster with integral load cell,” he continues. “The combination of the Biodex seating arrangement and our modified HapticMaster robot created the new ACT3D system.”

How It Works

“This is a robot that’s run by people. Subjects push on the robot load cell itself and the load cell senses the force that is exerted on the equipment and passes that information to the computer system,” Dewald said.

The computer creates a virtual world with objects and responds to movements that the patient makes, so the person’s arm is never forced into a particular position. The robot has 3-D information including the position in space where the subject interacts with its end effector, so it generates the sensation of contact with physical objects. The video interface allows the subject to see his or her virtual arm and the objects in space, and force feedback by the robot generates a realistic feel of the objects in the virtual environment.

The built-in interface supports the hand and forearm and is connected to a gimble. The gimble has the load cell connected to it that measures forces and torques. The tip of that robot is connected via the gimble and a rigid splint support to the forearm of the individual user.

Entering subject-specific limb length into the software allows the system to continually monitor the position of the elbow and shoulder, as well as provide a real-time display of the information.

“What is unique is that we can set up the environment to get the maximum benefit from the training,” Dewald said. “Over time, we can make subjects’ arms heavier to the point where they can deal with the real weight of their arms as they are exploring the workspace. It’s like two systems in one, first providing a virtual world that allows stroke subjects to move while also continually monitoring their advances in all therapies simultaneously. We have specific tests that go beyond the weight of the limb. We can make it seem twice or three times as heavy as it normally is—in essence, what it is like to move a heavy object in space.”

Increasing weight in a very gradual and incremental way simulates highly targeted strength training. Over time, subjects relearn to deal with the weight of the limb and then even more weight, to increase activity and widen their workspace. The applications programming interface lets the subjects see what they are doing on the video screen. A complete history of their progress is stored to compare changes.

“You have true numbers to compare. No guesswork. It’s heartening to the individual because the person can see the actual changes and it gives encouragement to go on and do even better,” Lam said.

Dewald added that subjects see not only the objects on the screen, but also a virtual arm that moves exactly as their arm moves. This avoids the problem of total virtual reality, which can sometimes be disorienting to people who may already have some visual problems related to a stroke. The virtual arm is in the exact configuration as their arm and moves in the exact way and at the exact speed as their own arm. So they are watching themselves in real time. It’s not just, ‘How do I get that little cursor to move up and down?’ They see their arm. They see the object. They feel the object. They know what they are doing in real time.”

What This Means for Therapists and Stroke Victims

This system will make the rehabilitation process much easier and more efficient, and remove guesswork. Therapists will have specific data analysis at their fingertips to see how well not just this robotic system, but all of their various therapies, are working with any individual.

In clinical trials, Dewald has found that chronic stroke victims improved their workspace by using this new robot for a total of 24 sessions over eight weeks. It is anticipated that acute stroke subjects will benefit even more. In short, the world generated by the robot indicates that people can get better even if their stroke happened 5–10 years ago and they have shown no progress for a long time.

“Are they going to be exactly the way they were before the stroke? Of course not,” Dewald says. “But, they can regain a lot of workspace, which means their quality of life will improve dramatically. They will be able to regain better control of things like closing doors, putting on clothes, and holding an object with their affected arm. Things that give people back their independence and dignity. That's what our robot, the ACT3D, is all about.”

Copyright ©2006 Medical Product Manufacturing News

Sign up for the QMED & MD+DI Daily newsletter.

You May Also Like