Sensory Feedback in Upper-Limb Prosthetic Systems
T. A. Rohland
Based on material which appeared in Myo-Electric Control Systems-Progress Report No. 12, University of New Brunswick Bio-Engineering Institute, Research Report 73.1, pp. 7-11, January 1973.
To Provide a Missing Element
The gradually increasing use of externally powered prostheses continually underscores the missing element in current systems-sensory feedback. Attempts have been and are being made to remedy this deficiency. With the permission of the author we present an abstract of material which originally appeared in Progress Report No. 12. Myo-Electric Control Systems, Bio-Engineering Institute, University of New Brunswick. Fredericton, N.B.. Canada. At the suggestion of the U.N.B, group we also obtained brief statements of experiences at Duke University and MIT. These statements are presented as Appendixes A and B following the U.N.B, article.
One of the main problems encountered in the use of present-day myoelec-trically controlled prostheses is the lack of sensory feedback. Even the small amount of feedback available in conventional cable-pull systems (socket pressure, harness pressure, movement, etc.) is absent when myoelectric systems are used. Thus the amputee must depend on his visual sense to position the limb and to grasp objects with the terminal device. This need for the amputee's full attention in order to control the prosthesis limits its usefulness.
Several researchers have tried various methods of feeding information back to the amputee, the most promising of these being the application of a vibratory stimulus to the skin. The purpose of the project reported here was to study different types of such stimulation, and to choose the optimum one compatible with the U.N.B, myoelectric control system; to build a feedback system employing this stimulus; and finally to make it easier for the amputee to control his prosthesis.
Choice of Stimulus
In approaching the problem it was decided to use electrical stimulation as opposed to mechanical vibration for the following reasons:
- The stimulus is applied through small electrodes rather than bulky vibrators which impose size limitations.
- There is less power drain on the batteries.
- There is less adaptation of the human body to electrical stimulation than there is to mechanical stimulation.
According to the results reported by several researchers1,3,4, the optimal signal characteristics for pain-free electrical stimulation are brief pulses (approximately 0.1-0.5 ms) combined in short trains, run at low pulse and train repetition rates, and delivered through relatively large electrodes.
Tests were performed to determine the type of modulation to be used in order to transmit the feedback information to the amputee. Both amplitude and frequency modulation were found to be acceptable but the phenomenon of "Phantom Sensation"5, using electrical stimulation, exhibited the inconsistencies reported by Gibson2, and was rejected.
An electrical stimulus applied to the skin through surface electrodes is picked up by the myoelectric control system and causes interference problems. This difficulty was overcome by incorporating a lowpass filter into the myoelectric amplifier and running the stimulus at a frequency higher than the myoelectric range, which is from approximately 20 Hz to 300 Hz.
Because of these considerations the stimulus finally decided upon consisted of 0.1 ms direct-current positive pulses from a constant voltage source at a frequency of 1000 p.p.s. This signal was delivered to the skin through silver/silver-chloride surface electrodes (using electrode paste) and the information was coded using amplitude modulation.
A decision was made to present pressure information to the amputee to enable him to grasp objects without watching to see whether he dropped or crushed them. The gripping pressure is sensed by two strain gages placed on cither edge of the thumb member of the prosthetic hand. The resulting signal is amplified and sent to a comparator which then adjusts the amplitude of the stimulus to the required level. This adjustment is done on a nonlinear scale such that approximately two-thirds of the range over which the stimulus amplitude varies is adjustable so that it can be set to cover the comfort range, from perception threshold to pain threshold, for each patient. The stimulus is delivered to the patient through a step-up pulse transformer so that high-voltage batteries are not required. The pulse transformer also serves as an isolation transformer for patient safety, eliminating any danger of continuous direct-current stimulation in the event of stimulation malfunction. The stimulating system is shown in block diagram form in Fig. 1 .
Tests and Results
To test the effectiveness of sensory feedback in giving the patient more control over his prosthesis, the system was tried on two amputees. The first was a 17-year-old girl who was a congenital below-elbow amputee and had been using the U.N.B. control system for two years. She tested the feedback system on an informal basis, picking up different types of objects (soft, hard, light, heavy, etc.), and found that sensory feedback was quite helpful. She liked the system and expressed a desire to have such a device incorporated into her own prosthesis even if it meant adding extra weight. The second amputee, a 13-year-old girl who, like the first, was a congenital below-elbow amputee, had never used a myoelectric control prosthesis before and therefore didn't perform the tasks as well as the first girl. She liked the feedback feature and expressed a desire to have one applied to her own prosthesis. The system was also tried informally by three occupational therapists, a prosthetist, and several engineers, all of whom expressed a keen interest in it and felt that such a system should be incorporated into the myoelectric system as soon as possible.
Formal tests were also performed with several normal subjects. They were asked to apply certain discrete pressures with the prosthetic hand to a pressure-measuring device while not watching the pressure gage. They did this without feedback, with feedback, and with their own (normal) hands. The mean square error for the series with feedback was approximately ten times smaller than that for the series without feedback, but two or three times larger than that for the series using the normal hand. Thus the feedback is seen to be measurably better in controlling a myoelectric prosthesis, although still not as good as human sensory feedback.
The results of the tests indicate that a prosthesis is easier to control when sensory feedback is provided and that the proposed technique is acceptable to the people using it as well as professional people knowledgeable in this field. Therefore, it seems highly desirable to provide sensory feedback for myoelectric control systems. The next project to be undertaken in this respect will be to redesign the system to use less power while providing the same functions, miniaturize it, and incorporate it into a prosthesis for amputee testing under normal conditions.
Bio-Engineering Institute University of New Brunswick, Fredericton, N.B., Canada
1. Gibson, R. H., Requirements for the use of electrical stimulation of the skin, in Clark, L. L. (Ed.), Proceedings of the International Congress on Technology and Blindness, American Foundation for the Blind, New York, 2:183-207, 1963,
2. Gibson, R. H., Perception of apparent movement from cutaneous electrical stimulation. Research Bulletin, American Foundation for the Blind, 9:13-21, April 1965.
3. Kato, I., et al. Human cognitional ability for electric stimulation signals. Proceedings of the Third International Symposium on External Control of Human Extremities, Belgrade, pp. 69-84, August 1969.
4. Kato, I., et al., Multifunctional myoelectric hand prosthesis with pressure sensory feedback system-Waseda hand 4P, loc. cit., pp. 155-170.
5. von Bekesy, G., Sensations on the skin similar to directional hearing, beats and harmonics of the ear. J. Acoust. Soc. Am., 29:4:489-501, April 1957.