Displaying Laterally Moving Tactile Information

Abstract.

1 Introduction

2 Perceptive and cognitive processing of tactile signals

In the skin of the fingertips, there are receptors responding to perpendicularindentation of the skin (SA I), to tangential displacement by friction (SA II), to vibration in the range of 5 to 40 Hz (RA), and to vibration in the range of 40 to 400 Hz (PC) [4].

In spite of the fact that scanning a rough surface should excite always SA I and SA II units and, depending on the surface pattern and the scanning speed, RAor PC units as well, it was shown that lowering the response of the SA II, RA,or PC units does not affect the perception of roughness [5, 6]. The respectivecognitive process seems to take only, or mainly, the SA I signals into account. Since the SA I units allow also a sufficient localization of the stimuli, tactile structures like braille can be read without any friction and within a widerange of scanning speed. Even more, the absense of friction as a source of "noise" should improve the reading performance [7].Although Braille readers commonly use several fingers of both hands in order to track the current and the following line and to explore in advance the structure of the text, e.g. the length of words and sentences, the actual reading is performed by just one finger of one or both hands each [8].

Navigating through a complex text is eased by dealing with a mental image of the text structure, and constructing such images is facilitated by some proprioceptive feedback reporting the position of the hand during the reading.

To summarize, "natural" tactile reading involves at least the stimulation of the SA I receptors by indentation, rather of several fingertips than of only one, and some proprioceptive feedback concerning the position of the reading fingers.

Vibrotactile displays offer an advantage mainly on the engineering side: The PC units can be stimulated by skin displacements of a few mm, i.e. by transmitting a small amount of energy. However, the user has to develop by training a pattern recognition process based on the PC signals. For this reason, results obtained by experiments based on vibrotactile displays apply not, or only partially, to braille reading and dealing with static tactile graphics.

The spatial resolution of the skin of the fingertips is different depending onwhether recognition, localization, or distinction of perpendicularly tapped ortangentially scanned tactile stimuli are concerned: A single point stimulus can be localized with an accuracy of about 1 mm. Two simultaneous tactile point stimuli are felt by tapping as two distinct objects if their distance is larger than about 2 mm. Structure widths of 0.1 mm are well perceived by scanning the object. Likewise, a movement is already recognized if a single stimulus is moving laterally for about 0.1 mm [9].Consequently, a braille dot or a raised line composed of several narrow dots or pins is felt as a single object when touched without tangential movement but its rough structure is perceived by scanning.

A slow sequence of lateral "jumps" of an tactile object can be felt as a smooth motion if the width of each jump is only a fraction of 1 mm.

As for the visual sense, one apparently moving object can be perceived if two stimuli are applied to distant points of the skin sequentially, but it feels different from an actually moving stimulus [10] - perhaps due to the high temporal resolution of the tactual sense.

Due to the adaptation of the SA I units, the structure of complex static information is well recognized only by scanning. Additionally, it is reasonable to assume that the determination of the structure is based not on the SA I signals itself but on their changes, i.e. on the movements of the edges of the indentations. In this case, tapping on a structure or popping up some information underthe resting finger provide a perceptual pattern that did not match any patternout of the temporal sequence obtained by scanning the same structure because the progressive indentation during the first phase of tapping causes only "leading" but no "trailing", i.e. releasing edges, and the edges move perpendicularlyto their own local tangent instead of unidirectionally on the whole fingertip as during scanning.

3 State-of-the-art devices

4 Providing the impression of moving dots

While the static spatial resolution of the display is given only by the partitioning of the surface, the dynamic spatial resolution, i.e. the number of stepsthe information is doing to move along a given distance, is determined by the distance between the surface parts and by the number of heights the parts of the surface can be driven to.

Laterally moving 8 dot braille characters results in 4 "dot tracks". Hence a dynamic braille display can consist of a solid surface with 4 rows of vertically moveable tactile pins. In the direction perpendicular to the movement, the width of the pins should just equal the width of the presented dots. In the direction of the movement, the width of the pins can be determined by the following considerations. During the movement of a dot for one pin width, the height of the dot should not change considerably in order to yield a nonvibratingdot. This can be achieved by holding one pin at full height as long as one neighbouring pin is lifted and the other neighbour is lowered. Consequently, one dot has to be presented by at least two or three pins depending on the phase of movement. Since there must be a gap between two dots in the same row, the maximal pin width plus spacing is a third of the lateral center-to-center distance of the braille dots.Graphic displays should comprise a pin array having the same spacing in x- andy- direction. On principle, presenting graphics using a frictionless display asdescribed could lead to a difficulty: a long horizontal or vertical stroke would perceptually vanish during a scanning movement parallel to its axis because the contents of the display would not change. Of course, the display controllercould recognize such situations and introduce small movements of the display contents perpendicular to the main movement. However, having a sufficiently highdynamic spatial resolution, this problem will virtually never arise because, as during scanning real structures, the user will never carry out exactly straight movements. As long as the amplitude of the wavy line he moves along is larger than half the spatial quantization caused by the dynamic resolution, the display contents will vary and the stroke will stay perceivable.Both braille and graphic displays of the kind described above have a peculiarity: With the fingertips resting on the moving display, the scanned objects are felt as smooth and the tactile pins itself are not distinguishable. However, ifthe user starts to scan the display surface the smooth objects turn into roughones because the gaps between the pins becomes perceivable although not localizable.

5 The first laboratory displays ("Braille Movies")

In the second version, the 11 inner PCBs of the stack are replaced by two thinner PCBs each resulting in a decomposable stack of 12 modules. Each module carries also the electronic drivers of the piezoelectric benders and a shift register. Both the volume and the profile of the display including these drivers match approximately that of an assembly of 4 usual braille cells.

The capacitances of the piezoelectric benders are not switched to one out of two DC supply voltages by means of open collector or push-pull output stages, asusually. Instead, in order to achieve a low power dissipation within the drivers even during fast movements of the displayed information, the electrodes are charged from a sinusshaped AC voltage of up to 500 Hz by means of controlled halfbridge rectifiers. Given a reading speed of 14 characters per second, a carriage return time of 1 second, characters with 3 dots on an average, a capacitance of 40 nF per bending strip, and a control voltage range of 200 V, we had an average heat dissipation within the 72 compactly arranged drivers of about 6 W with push- pull outputs but we have less than 0.1 W with controlled rectifiers. Each electrode can be charged to an arbitrary value within the voltage range of the AC voltage by changing the state of the corresponding driver each time the AC voltage reaches the desired value. So, an arbitrary force within the given range can be exerted on each individual tactile pin.

A microcontroller on an additional PCB is executing the following tasks: Communications with the attached personal computer using a serial interface, determining the position of the display unit and hence the section of the virtual braille line to be displayed, conversion of the supply voltage of 5 V DC to 200 V DC and to 200 Vpp AC by means of switching converters, and controlling the rectifiers which are driving the bending strips. So, except for the geometry of the display, the way the text is displayed may be altered rather easily by softwarechanges.

6 First experiences testing the Braille Movie

a) There was some tactile noise in the background, i.e. though the display should be nonvibrating sometimes some pins were actually vibrating.

b) Neighbouring dots appeared not distinctly separated from each other, i.e. a"C" could be taken for a horizontal stroke, for example.

c) Braille dots 2 and 5 were perceived weaker than the rest of the dots.

Topic a) was due to a software bug we could not fix before the beginning of the exhibition. Of course, this is not an inherent problem.

b) should be solved or at least considerably improved by changing the shape ofthe pins. Using pins with more narrow ends in the direction of the informationmovement, the gaps between neighbouring dots will be better perceivable. Furthermore, as mentioned in section 4, the horizontal distance of the braille dots can be enlarged. The consequently enlarged width of the virtual line can be compensated by moving the virtual line contrary to the direction of the movement of the display unit [2].

It seems rather unlikely that problem c) was caused by generally weaker bending strips in the second row. Perhaps this is a general effect with this kind of displays. We have enlarged the row- to-row distance to 3 mm in the second display in order to avoid dots 2 and 5 to be hidden by the other dots.

Having in mind that this was the very first approach and that the display is certainly improvable, we feel that the basic principle will become a sufficiently convenient way to display Braille when costs or size and weight of the outputdevice are critical factors.

7 Possible features of devices containing a Braille Movie

Generally, some coarse tactile elements on the right of the high resolution display would allow a prerecognition of the word boundaries. Besides functions supporting the navigation and orientation in structured texts and tables [2], features concerned with reading plain texts could be valuable, too. For example, afast movement of the display to the left end of the line could cause an automatic line feed. In this way, a large text could be read just by moving the display.

8 Future research topics within the Braille Movie approach

The reading performance of both skilled and unskilled braille readers should be compared using paper braille, conventional braille lines, and a Braille Moviewith combinations of the mentioned display modi.

Learning braille using a Braille Movie from the beginning could be advantageous because it may help to concentrate on just one fingertip, and since the computer can track the reading finger a teaching system could provide some individual help by speech.

Also, it would be interesting to display ink print characters and plain graphical elements using a "Graphic Movie". Compared to vibrotactile displays, we expect the users to show a similar performance but to become skilled after a considerably shorter period of training.

References

Home