The engineer who trains rehab robots –

0
56

Armin is standing directly on the window front of the open-plan office. This sturdy, green-silver machine with the long, articulated gripper arm looks as if it came from a factory hall for assembling cars. But it is located in a multi-storey research building opposite the Inselspital Bern.

"This is our therapy robot," says Laura Marchal-Crespo, slipping her arm through the loops for the demonstration. A few years ago, Armin was developed at the ETH Zurich by Robert Riener and Tobias Nef to train the hand and arm movements of patients with paralysis after a stroke. Now the 39-year-old professor at the Competence Center of the University of Berne for Biomedical Engineering (Artorg) is working on adapting the control of the rehabilitation robot to individual needs.

Every year, around 16,000 people in Switzerland suffer a stroke. Those who survive such a cerebral infarction often have to contend with persistent failure of the motor system. Worldwide, the disease is one of the most common causes of paralysis. Intensive physical and occupational therapy can often bring back a lot of flexibility. Nevertheless, such a treatment is usually very time and effort consuming. Because the patients must be supported or guided to re-learn the gripping and walking again.

"The longer a patient is inactive, the harder it is for him to regain mobility later," says Marchal-Crespo. It is also important to repeat the exercises constantly so that the brain structures can reorganize themselves. However, rehabilitation robots are by no means intended as a replacement for conventional therapies, but as a useful supplement.

To learn from mistakes

The Spaniard Laura Marchal-Crespo was already researching at ETH Zurich at the interface between man and machine with the help of robot assistance and virtual reality. She was particularly interested in helping paralyzed people after a stroke. Before the ETH she studied and obtained her doctorate at the University of California in Irvine. In 2017, she was appointed professor of the Swiss National Science Foundation (SNSF) and has since worked at the University of Bern. This October she held at the local faculty for medicine her inaugural lecture.

During training with Armin, the impaired arm of the patient is strapped to the exoskeleton of the movable robotic arm and follows the movements dictated by the robot. In this way everyday situations in front of the screen, such as pouring water into a glass, can be practiced again and again.

"We improve the ability to move by reinforcing the mistakes," says the engineer. This would sound strange. However, if a stroke patient, for example, wants to grab something, but does not make it precise, the rehab robot reinforces this inaccuracy by creating a slight resistance. In this way, the patient has to work harder. The learning algorithm developed by Marchal-Crespo independently decides which patient complicates the task and which one makes it easier.


During training, the robot decides whether to help or resist. Photo: Adrian Moser

In addition to Armin, there is also a horizontal running machine in the Bern open-plan office. Marchal-Crespo can then examine a patient even during the exercise in a magnetic resonance tomograph. Lying, he either actively or passively moves his legs there using a knee-jack robot. The researcher wants to find out which areas of the brain are stimulated by such specific exercises and how the neurons re-connect after a stroke. She is currently trying to simulate the feeling of walking as real as possible, so that the pressure on the soles is felt.

Originally she came from the suburbs of Barcelona, ​​she laughs. She was very fortunate that she received a scholarship for the USA at the time. Because her parents were not academics, and her father earned money as a taxi driver. Her mother has always supported her and her two sisters in every way, she adds. One is now a mechanical engineer in Australia, the other a chemist in Barcelona.

Because Marchal-Crespo's man comes from the capital of Catalonia, he speaks with the three children together, aged three to seven Catalan, while she speaks Spanish with them. Everyday life with family and work is a challenge, but it is also exciting and a privilege, she finds. Her husband works as an engineer at ABB in Zurich. As the family lives there, she commutes to Bern four days a week. "If it's possible, we'll all do outdoor activities together on the weekend and either go hiking or cycling."

Therapy with Virtual Reality

Marchal-Crespo is fascinated in her work above all by the sheer inexhaustible possibilities of the virtual world. So one of her doctoral students now brings out a VR glasses. If you open it, you are in a virtual room where suddenly an apple, then a pear or another apple is hanging in the air. With a controller you now have to control a ball exactly to these fruits and at the same time count the fruits.

"It sounds like it's very easy," says the scientist. But for patients after a stroke, the combination of motor activity and a cognitive task is difficult. Because they have to focus heavily on both and only regain these skills.

Also use games

Marchal-Crespo has other gadgets in her test lab. For example, a special joystick that allows you to drive a small ball over various rough, smooth, soft, hilly or canyon-like surfaces and feel the particular material on the hand you just touch in the virtual world. "We try to activate as many senses as possible in order to achieve the greatest possible training success," she says.

Their goal is to motivate patients to exercise longer and more intensively. Of course, this is exhausting and often frustrating, she emphasizes. When a child learns to walk, drop it, but get up again, try again and again. But that comes from the inner drive. Those who have to practice walking again after a stroke need much more strength and stamina.

She is currently investigating whether she can increase patient motivation with special games. "I also want to make boring training sessions exciting," says the researcher.

Created: 09.11.2019, 21:04 clock

LEAVE A REPLY

Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.