Tactile display on the remaining hand for unilateral hand amputees

1 Introduction

With versatile sensory and motor capacities, hands are the most dexterous body parts to interact with the environment. Hand amputation, mostly caused by trauma events or diseases, causes lots of inconvenience into everyday life. Commercial hand prostheses partially restore the hand motor functions, but in most cases not the sensory abilities [1]. Academic research on tactile feedback for hand prostheses has provided various solutions. With the recent breakthroughs, invasive tactile feedback approaches to provide tactile feedback [2], [3] would provide the best results in the long run. With less risks, noninvasive approaches [4] are expected to benefit hand prostheses users in the near future.

Previous research has shown that noninvasively displaying tactile information generated at the five fingertips of a hand prosthesis by a miniature tactile display device is not trivial to achieve. Except on hands and faces, the two point discrimination distance on human body surface is mostly beyond 3 cm [5]. Therefore, small areas such as the upper arm is in principle not sufficient to noninvasively display the information of five distinct locations. Another challenge is loss of reference: subjects have frequently reported to lose tracking of tactile feedback locations [6]. Applying tactile feedback to a larger body area, such as the back or around the waist, shows good results of perception, but the devices are bulky and awkward to use.

We propose in this study to use the remaining healthy hand as the location to apply tactile feedback for prostheses used by unilateral hand amputees. Our hypotheses about this tactile display approach are: distinguishing five tactile feedback locations would be easy to achieve with little or even no training; confusion among fingers could initially be a challenge, but will be overcome during the device usage. Although there have been tactile feedback gloves available on the market, these gloves mostly provide feedback on fingertips and/or the palm. Tactile gloves designed with this approach would interfere with the normal sensory and motor functions of the healthy hand when used as a tactile feedback device for amputees. In this study, we designed a tactile feedback glove which provides feedback on the back of fingers to reduce influences.

In what follows, our proposed tactile feedback concept and the designed glove is presented first. Then experiments designed to evaluate the effectiveness of the tactile feedback device are described. Finally, we analyze the results and discuss the implications.

2 Tactile feedback device

Although the dorsal sides of fingers are not as sensitive as the ventral, localizing on which finger a physical stimulation is applied is still quite intuitive. To deliver the tactile information of a hand prosthesis for unilateral hand amputees, a tactile feedback glove is fabricated and to be worn on the remaining healthy hand. The glove delivers tactile information generated on fingertips of a hand prosthesis to the back of corresponding healthy fingers of the remaining healthy hand. The main design objectives are: intuitive tactile information display, a compact size, light weight, and easy to wear. The tactile feedback design concept and designed glove are shown in Figure 1.

Figure 1

Tactile feedback concept. A tactile display glove is worn on the remaining right healthy hand by a unilateral hand amputee to display tactile events generated on the fingertips of the hand prosthesis (left). The tactile display actuators (pancake vibrators, indicated by white dash circles) are fixed on the glove and placed at the back of each finger.

The tactile feedback glove (shown in Figure 1) mainly consists of a sport glove, five eccentric rotating mass (ERM) vibrators (iNeed Inc., HK), and control electronics. A sport glove fits tightly to the fingers, which is essential to reliably deliver vibrotactile information to the back of fingers. A sport glove is also easy to take on and off, which is an important consideration to facilitate the independent usage by hand amputees. The ERM vibrators have the advantage of low power consumption, compact sizes (diameter 10 mm, thickness 3.4 mm), low costs, and an easy control interface. These vibrators provide tactile feedback at the back of each fingers to minimize the influence to the motor and sensory functions of the healthy hand.

3 Material and methods

To preliminarily evaluate the effectiveness of the tactile feedback concept described in the previous section, we designed experiments to test the tactile feedback glove with healthy subjects. These experiments have been approved by the institutional ethics committee of the Lund University in Sweden.

3.2 Experiment setting

In preparing an experiment, a test subject sits on a chair and rests her/his right arm on a table in front of her/him. The tactile feedback glove is worn on the subject’s right hand and adjusted to fit the hand well. During an experiment, a test subject is asked to move his/her right hand around while perceiving tactile information applied on the back of fingers to mimic our intended application scenarios. Hand movements increase the difficulty to identify the tactile feedback applied by the glove. An investigator issues stimulation signals by a piece of control software running on a laptop. The software interface is shown in Figure 2. The subject reports verbally his/her judgments about the tactile feedback to the investigator. The investigator records the answers on data record forms.

Figure 2

The software interface used to issue stimulations during experiments. An investigator gives tactile feedback signals by clicking on one of the three buttons on each finger. The three buttons correspond to three feedback levels: strong (S), medium (M), and weak (W). The activated feedback is indicated both by a blue color on the button and the status panel.

4 Results

Experiment results are presented in this section as a preliminary evaluation of the tactile feedback concept proposed in this study.

4.1 Finger detection

The error rate and adjusted response time (excluding stimulation duration) of distinguishing tactile feedback stimulations applied on the back of five fingers without training is shown in Figure 3. One can observe that in general both the error rate and response time decrease as the stimulation duration increases. With a stimulation duration of 100 ms, subjects have managed to distinguish the majority of displayed finger localization information correctly.

Figure 3

The error rate and response time (solid error bar) to localize tactile display on the back of fingers. The error rate is high when applying a short stimulation duration of 50 ms and the error rate reduces when the stimulation duration is larger than 100 ms. The response time tends to decrease as the stimulation duration increases.

4.2 Force level detection

The experiment results of test subjects using the tactile feedback glove to perceive different levels of stimulation is shown in Figure 4. Because the subjects were tested without a training session in our study, they needed to learn the relative stimulation strength during experiments. It was observed that the medium feedback level was most frequently misjudged.

Figure 4

The error rate (A) and confusion matrix (B) of distinguishing three stimulation levels displayed on the back of the index finger. The medium stimulation level is the most difficult to recognize. It tends to be mistakenly recognized as the strong feedback. Note that the confusion matrix includes data from all five subjects, so there are 25 tests for each stimulation level.

4.3 Learning effect

Although displaying tactile information on the back of fingers does not feel as natural as on fingertips, response time shows subjects gradually get used to this tactile information display approach. As shown in Figure 5, subjects’ response times tend to decrease as the experiments proceed. We also observed during the second experiment session that mistakes for force level detection happened mostly at the beginning. Moreover, test subjects also reported that they could perceive the information transferred by stimulations easier over time. However, the result tends to be biased by the stimulation duration: durations of 500 ms and 1000 ms also have the shortest response time.

Figure 5

Learning effect observed during the six test groups of experiment session one. Test subjects’ response time generally decreases during the session as the experiment proceeds. The stimulation durations of test groups #1 to #6 are: 250 ms, 100 ms, 80 ms, 1000 ms, 50 ms, and 500 ms, respectively. The increased response time of tests #5 and #6 can be explained by the difficulty to detect the ERM stimulation duration of 50 ms.

5 Discussion and conclusion

Our preliminary experiments on five healthy subjects show that displaying tactile information on the back of fingers is feasible and intuitive. The displayed information (finger localization and contact force level) can be readily recognized even without training.

Even though it seems to be disturbing to display tactile events occurred at fingertips of the left hand at the back of the right hand, our study shows that subjects learn the tactile display quickly. The adaptability of human body surface to new tactile information has also been observed in previous studies [7].

The tactile feedback concept proposed in this study is to be evaluated with amputee subjects. We expect that this tactile display method will work well with unilateral hand amputees. When a hand amputee subject manipulates objects with a hand prosthesis, the tactile events are generated actively. As a result, the amputee has a prediction of the tactile feedback before it happens. Therefore, an actively generated tactile event is probably even easier to recognize than a passive one.