The University of Southampton

Iterative learning control for re-education of upper limb

Robotics and Control

When you practice playing tennis you become better at it, because new nerve connections have been made within your brain and spinal cord. Not only do you need to practice, but you also need feedback of your performance so that you can correct your movement. In this research we are using this idea to teach people who have had a stroke how to learn new skills.

A Stroke is usually caused when a blood clot blocks a blood vessel in the brain. It acts like a dam stopping the blood reaching the brain downstream. As a result some of the connecting nerve fibres die and the person becomes partially paralysis on one side of the body, this is called hemiplegia. These fibres cannot re-grow, but the brain has plenty spare capacity so new connections can be made. In fact the brain is continually and rapidly changing as we learn new skills; new connections are formed, redundant ones disappear. When people re-learn skills after a stroke they go through the same process as you do when you learn to play tennis. But they have a problem because they can hardly move at all so they cannot practice which means they don’t get feedback. Muscles can be made to work by Electrical Stimulation. Electrical impulses travel along the nerves in much the same way as the electrical impulses from your brain. If stimulation is carefully controlled, a useful movement can be made. This works better if the person is attempting the movement themselves; we therefore need to combine a person’s own effort with just enough extra electrical stimulation to achieve the movement. This is what we will do in this project by adjusting the level of stimulation in response to the person’s movement.

To teach people who have had a stroke how to move their arm we will ask them to track a spot of light by moving a vertical rod over a flat board, like moving a chess piece. As they move we stimulate their muscles. If they track the target well, then on the next attempt we turn the stimulation down if not we increase it. To get the level and the timing of the stimulation right we measure the difference between the direction of the movement of the arm and the movement of the spot of light. We then adjust the stimulation in a way that we think will reduce the difference, ideally we want them to follow the same path exactly. After the person has had another go at following the spot of light we measure the difference again and make the adjustment again. In fact each time we make an adjustment to the stimulation we measure the effect so that we can continually improve their accuracy. This is called Iterative Learning. Adjustments are made according to a set of rules - making these rules is an important part of the project. This process is very similar to the way your brain works when you are learning to play tennis. One important addition though is that we need to keep reducing the stimulation to encourage the person to use their own effort to follow the spot of light, rather than relying on the stimulation. So if they track the spot well, then the next time they get less help from the stimulation. This technique of iterative learning is often used in ‘training’ robots for industrial purposes, but as far as we know nobody has tried using it to help people who have had a stroke learn to move again.

Primary investigators

Associated research groups

  • Electronics and Electrical Engineering
  • Information: Signals, Images, Systems Research Group
  • Electronic Systems and Devices Group
  • Electrical Power Engineering
Share this project FacebookTwitterWeibo