Control of functional electrical stimulation extended physiological proprioception C. Kirtley

and B.J.

Bioengineering

with

Andrews

Unit, University

of Strathclyde,

Glasgow

G4 ONW, UK

ABSTRACT The use ofjitnctional electrical stimulation (FES) of muscle for paraplegic locomotion, or grasp augmentation in tetraplegia, is limited by the variability in muscle response to stimulation as a result of several external and internal factors. Previous approaches to this problem have used position-servo controllers, which have been shown to finction satisfactorily in the laboratory. However, such systems willfail should obstacles be encountered or should the stimulation hardware develop a fault. Toprevent suchpotentially dangerous failures some form of sensovyfeedback is required. This paper describes thefirst application of a technique known as extendedphysiologicalproprioception (EPP) to the control of FES to compensate for muscle response variability and provide proprioceptive feedback via the appropriate sensory pathways. In the experimental system described, a paraplegic subject controlled the extension of his paralysed knee by shoulder protraction. A Bowa’en cable linked the two joints, and a dynamometer in this cable was used to derive the control signalfor a computer-controlled stimulator which delivered surface stimulation to the quadriceps muscle group. Modelling and parameter io!entiJcation were ptiormed by analysis of the step response, and the controller was designed from consideration of the root locus. The advantages of the system, in terms of improved proprioceptive feedback and reduced limb-positioning error were assessed in a test ofjoint positioning accuracy with vision occluded. The EPPsystem showed improvements over both open and closed-loop position-servo controllers. Keywords: feedback

Funtional electrical stimulation, paraplegia,

INTRODUCTION Functional electrical stimulation (FES) of paralysed muscle has enabled standing and walking of paraplegics in the laboratory’“, and grasp in tetraplegia4. However, for these systems to be practical for use in the home and street environment, a number of problems related to the control of the stimulation need to be addressed5. First, the force output (and hence joint position achieved) by stimulation of a muscle varies over a considerable range as a result of many internal and external influences, such as temperature, fatigue, and relative movement between the stimulating electrode and the motor point of the muscle5. One solution to such a problem is to implement closed-loop positionservo control, in which the joint osition is monitored by an electro-goniometer an B compared with a reference to give the controller input. The reference signal to the controller may be obtained from a potentiometer moved by a normal joint7y8, EMG from a normal muscle7, or from a predefined sequence stored in the computerg. Position-servo control methods have been shown to function satisfactorily in laboratory conditions, in which the paralysed ‘oint is free to move. However, practical systems wil 1’ need to negotiate obstacles in the path of the limb, and it is therefore necessary that there be some means by which the user can detect these obstacles and modify the stimulation accordCorrespondence

and reprint requests to: Dr C. Kirtley

control, extended

physiological proprioception,

sensory

ingly. In simple systems employing hand-switching for control, this is possible to the extent that the timing of muscle contraction can be controlled, but not the amplitude of the force developed. Moreover, the coordination of the timing of switch pressing places a large cognitive load on the patient. Hence it is desirable that the atient be given continuous control over the stimu Pation in such a way that the information flows more naturally from nervous system to orthosis. In laboratory systems so far im lemented, the sensory feedback function has usu 2 ly been implemented by relying on the patient’s vision. This is far from satisfactory because it demands the continuous attention of the patient. It has been suggested that supplementary sensory feedback (in the form of electrocutaneous stimulation) might be employed to lessen this reliance on vision’. However, many such systems have been tried in conjunction with both upper and lower limb prostheses’~‘2, with disappointing results in terms of practical impacQ3. This may be because the sensory information is transmitted along inappropriate nervous pathways, which are not part of the body’s normal sensorimotor systemi4. Thus the information must be presented in a form which can be acce ted, and skin vibration or pain receptors are unlike P to make a satisfactory substitute for angle receptors IY. A better approach is to feed back joint angle information directly to the normally innervated jointi6. This principle is called extended physiological proprioception (EPP) or control interface

0 1990 Butterworths for BES 0141-542.5/90/030183-06 J. Biomed. Eng. 1990, Vol. 12, May

183

Control ofFES: C. Kirtlq and BJ Andrews

feedback, and has been successfully applied to the control of several upper-limb prostheses15-1g. By constraining the motion of the control joint according to that of the slave joint, a consistent relationship between force, position and velocity of the two joints is created: the so-called ‘unbeatable servo’. This constraint can be most easily effected b forming a mechanical linkage between control an (Y slave joint, for example, by means of a Bowden cable. The tension in this cable can then be used to control the prosthetic or orthotic actuator. In this way, proprioceptive sensations arising from the slave limb are transmitted to the control joint in a physiological manner, via the normal proprioceptive pathways”. The purpose of our study was to investigate the possibility of controlling electrically stimulated muscle by EPP, with a view to applying such a technique to the control of FES-powered neuroprostheses. This

Bowden

cable)

/

Figure 1 Experimental FE.5

184

J. Biomed.

apparatus

for investigation

Eng. 1990, Vol. 12, May

of EPP control

of

paper discusses the basic problems encountered in attempting to control FES by EPP, and includes an assessment of the method compared with open-loop and closed-loop position servos.

MATERIALS

AND METHODS

To develop and test the concept of controlling stimulated muscle by EPP, the experimental apparatus shown in Figure 1 was designed, in which a araplegic subject controls his paralysed knee joint FJy FES applied to the quadriceps muscle group. Although in this case a lower-limb joint was controlled b upper-limb musculature, the arrangement model for EPP provide dy a suitable experimental control of FES in general. A tram-scapular harness rostheses similar to that used to control upper-limb was worn, and a nylon-in-steel Bowden cab Pe (Perlon) linked shoulder protraction to knee-joint extension by its attachment to a pulley on the joint of a knee brace. The arrangement was such that shoulder retraction caused tension in the cable to rise and the K, ee tended to extend. Actual extension was difficult without stimulation because of the small mechanical advantage given by the pulley radius, which was 25mm. Tension in the cable was monitored by a strain gauged proof-ring dynamo-meter in series with the cable at the knee. The amplified signal was input to the analogue/ digital (A/D converter; 12-bit, input range -5 to +5 V, Type PC-26A, Amplicon Liveline Ltd, Brighton, England) of an IBM-PC compatible microcomputer, and used to generate the control signal to a constant-current stimulator develo ed at this site. The stimulation train was pulse-wid tK modulated at a constant inter-pulse-interval (40 ms), between the threshold level of the muscle (typically 50-100 ps) and a limit, arbitrarily set at 5OOps, and applied using Pals + surface electrodes (Axelgaard Manufacturing Co. Ltd, Fallbrook, CA, USA). Current level was adjusted to give full extension at this limit, and the force transducer was calibrated for cable tensions produced when comfortable levels of force were applied through the harness. A block diagram of the system is shown in Figure 2, which is modified from an EPP control system used for an upper-limb prosthesislg. The force sensed by the dynamometer is denoted by T.,,(+ which is a function of the difference between the user’s intended and the actual position of the position, &[m(t)], control joint, x(t). In a single cable system such as this, in which the action of gravity is used to rovide a restoring antagonistic force, a 1: 1 re Pationship between the control and paralysed joints will exist only whilst the cable remains taught. It is, of course, possible for the cable to become slack should the limb move faster than the control joint, in which case the advantages of the linkage would be roprioceptive Post. In practice this limitation of single cable s stems is not a serious one, because the user can feel x at the cable has become slack and, by moving the control joint accordingly, can re-establish the linkage.

ControlofFES: C. Kirtley and BJ Andwws Normal

[control)

joint

Paralysed

(slave)

-4

I-

joint

Feedback

Control of functional electrical stimulation with extended physiological proprioception.

The use of functional electrical stimulation (FES) of muscle for paraplegic locomotion, or grasp augmentation in tetraplegia, is limited by the variab...
710KB Sizes 0 Downloads 0 Views