Robotic devices in control during sleep

It has actually been common knowledge for a long time that if you rock a baby in a cradle, it will fall asleep more easily. Regular movements, for instance in a hammock or on a moving train, help adults to get to sleep as well. But exactly how these movements work, and whether they could help people with sleep disorders has scarcely been investigated up to now. Robert Riener, Professor for Sen?sory-Motor Systems, intends to fill this gap. Together with Peter Achermann, who heads the Group for Chronobiology and Sleep Research at the University of Zurich, he has de?veloped special beds to examine the relationship between motion and sleep. One bed makes a gentle pendular move?ment, like a cradle. The other pushes the sleeping person to and fro horizontally, either crossways or lengthways, and rises and falls as required. There is a reason for clearly separating the movements, explains Riener, ¡°If we mix the three patterns, then it¡¯s more difficult for us to examine their impact. And the test persons become nauseous more quickly.¡±

Soft and silent

Right now, the beds are in the test phase. The test persons are fitted with measurement sensors in the sleep lab, which provide information on the depth of sleep. Besides brain waves the researchers also measure respiratory rate and heart rate. Using the data recorded, they wish to establish how the movements affect sleep onset behaviour and sleep quality. The next step involves the bed being directly controlled by the physiological data. As soon as the device notices, based on brain waves and heart rates, that the test person is sleeping less deeply, it endeavours to prevent from that person waking up by means of corresponding movements. If it succeeds, this could help people with sleep disorders to achieve deeper and more refreshing sleep. Reiner can also envisage other applications: ¡°Movements of this kind during sleep could also prove beneficial for people with psychological disorders.¡±

For Riener, who as an engineer has been involved for years in the development of medical devices, there are, however, several technical questions. For instance, the construction of the bed constitutes challenges that should not be underestimated. The motors have to move the bed ¨C that together with the test person weighs more than 100 kilos ¨C without the slightest jolt. Furthermore, it must be practically noiseless to avoid sleep being disturbed by side-effects. It will be even more demanding when it comes to linking the mechanics to human physiology. ¡°The depth of sleep is not a parameter which one can simply determine using a set value¡±, says Riener.

Adjusting specifically to the individual plays a key role in another project of this ETH engineer. Together with his group he is developing an active prosthesis by means of which patients whose leg has been amputated can carry out movements which simulate those of the healthy leg. ¡°Most prostheses are passive substitute organs¡±, explains Riener. ¡°Our model is different. The artificial knee is equipped with a motor that moves the artificial lower leg.¡±

To enable the movements to mimic the natural model more closely, the joint also has mechanical springs which support the motor and protect against sudden jolts. With the active prosthesis, the wearer can climb stairs more easily or walk on uneven ground. And the prosthesis can do something else too. It is equipped with pressure sensors which allow the patient to feel how heavily he is treading on the ground by means of electrodes fitted to his back.

Man must retain control

According to Riener, the biggest challenge for this device is interaction with the wearer. ¡°The prosthesis should automatically recognise the intentions of patients to enable them to really use it like a healthy leg.¡± Right now, control is exercised via sensors attached to the clothing of the healthy leg. Based on these sensors, the prosthesis recognises what movement the patient wishes to make. This automatic adjustment is only possible because the researchers use sophisticated control algorithms which drive the motor quickly and flexibly. The important thing is for the human to retain control. ¡°The prosthesis must adapt to the man and not vice versa, otherwise the patient will not accept it¡±, insists Riener.

The knee prosthesis is also currently in the test phase. ¡°Our device is actually only intended for a limited target group, for older patients to be exact¡±, explains Riener. Young, strong patients found it far easier to climb stairs and didn¡¯t necessarily need a device of this kind. The example of the active prosthesis demonstrates very clearly that the successful development of such devices is dependent on close cooperation between engineers and doctors, nursing staff and patients. ¡°As an engineer I have to know what problems patients are confronted with in their daily lives and where there is a need for mechanical support¡±, comments Riener. ¡°Otherwise, we will be developing devices that nobody needs.¡±

Roger Gassert, Assistant Professor of Rehabilitation En?gineering, has also positioned his projects at the interface between the engineering sciences and medicine. Like Riener, who developed the motion robots ¡°Armin¡± and ¡°Lokomat¡±, innovative devices for stroke patients, Gassert looks at novel approaches to rehabilitation too. In one of his projects, he consciously focuses on hands. ¡°There are already numerous devices which support the movements of the shoulders, elbows and wrists¡±, explains the engineer. ¡°But what are still missing to a large degree are robotic devices that train hand functions.¡± These would, however, be particularly important for stroke patients as they normally find it difficult to open their hands and grasp objects. Often, their sensory skills are also impaired along with their motor skills. In this case, patients can no longer really feel how strongly they are grasping an object.

Gassert and his team have developed a robot which promotes the exchange between hand and brain in both directions. In one exercise, blindfolded patients have to pick up wooden bricks of various lengths with their fingers and then recognise which brick they are holding, based on the hand position. The robot developed by Gassert mimics this exercise. The patients pick up the virtual bricks by pressing two movable grips against each other. The robot defines via the control how long the grips can be moved freely, and when they have to be blocked. The patient then feels how big the virtual bricks are.

With the robot, sensory skills can be measured objectively. Gassert is convinced that ¡°In their day-to-day practice the therapists have a keen sense of whether someone is making progress. But if we want to move forward in this field, we need more objective measurement data.¡± At the present time, the device is being tested in a rehabilitation clinic in Canton Ticino. The first results are encouraging: the work with the therapy robot seems to lead to a substantial improvement, not least because with this device the intensity of the exercise can be increased gradually.

Incise, prick, stitch

A second project, on which Gassert is working intensively right now, has a completely different objective. In collaboration with researchers from other universities, he is in an EU project to develop a surgical robot which should be capable at a later stage of carrying out certain simple operations. The goal is to relieve the burden on doctors. ¡°Similar to the control of an airplane, a human would in future primarily assume the more complex phases, whereas for the simpler ones he would simply have to monitor the machine¡±, explains Gassert. But the scientists are still a long way from this. Right now they are constructing the first prototype. If it functions as planned, it will be able to carry out three tasks: remove a tissue sample by puncture, stitch up a small wound and excise tissue from a wound. The robotic device is controlled using data that are collected before and during the operation, using imaging techniques. This enables the machine to react if organs shift during the operation. In concrete terms, the surgical robot consists of two movable ¡°hands¡±, each having two ¡°fingers¡± to grasp something. The hands are fixed via two movable ¡°arms¡± to a rotatable board which is attached, in turn, to a rail system. ¡°The construct mimics a doctor who is bending over the patient¡±, explains Gassert. In total, the device has 18 degrees of freedom ¨C so the control require?ments are correspondingly high.

Feeling the force correctly

One major innovation is that the device is fitted with force sensors. During the operation the robot can recognise the force being exerted on the tissue. This means it is possible to avoid damage to the organs. When a doctor is stitching up a wound, it is good to be able to measure the force too. In the case of today¡¯s surgical robots, the doctor can see on the screen that the thread is taught, but not how much force is being applied to it.

The various parts of the robot, developed at numerous universities, were assembled in Gassert¡¯s laboratory this autumn. ¡°Here in Zurich we have concentrated on the development and control of the robot¡±, he reports. ¡°The other teams developed data collection and monitoring.¡± If everything goes according to plan, the engineers will soon be practising the first operation steps on animal organs with doctors at University Hospital Zurich. ¡°It will be a few years before the robots can carry out operations independently¡±, comments Gasser dampening any overly high expectations. ¡°Amongst other things we have to clarify a few safety issues, such as what happens to the patient if an unexpected error occurs during the operation.¡±