Display of Virtual Braille Dots by Lateral Skin Deformation: Feasibility Study

3. Parameter Tuning

Braille is normally produced according to strict geometrical specifications, but the manner in which these specifications translated into the VBD's parameters was not straightforward. For this reason, a first experiment was carried out to find the parameters that produced virtual Braille of appropriate dimensions. For the purposes of the feasibility study, only the width and the separation of dots were adjusted. The amplitude of the virtual dot sinusoid was set to the maximum that the system could provide. The amplitude and the wavelength of the texture were set empirically to values equal to 1/8^th those of the virtual dot sinusoid.

3.1 Method

With the help of the reference subject, the width of a virtual dot was first adjusted to match the sensation caused by a single physical Braille dot. Then, the distance between dots within a virtual character was adjusted. The inter-character spacing value was inferred from the value found for intra-character dot spacing.

The tuning experiment was conducted following a 2-alternative forced choice protocol, using the two-hand method in order to speed up the process and facilitate comparisons. The subject was asked to touch a reference stimulus produced by a conventional refreshable Braille display with her left index. She then immediately explored two stimuli on the VBD with her right index and selected the one that best matched the reference stimulus. Dots were always displayed with texture. After a short experimentation used to determine an appropriate range, the virtual dot width was varied among six equally spaced values from 0.5 mm to 3.0 mm. The intra-character dot spacing of the character 'c' ( raised raised ) was similarly varied from 1.0 mm to 2.5 mm. The dot width found in the first step was used in the second.

3.2 Procedure

Both tuning experiments proceeded in the same manner. The reference subject moved the VBD toward the left, waited for an audible signal, explored the two different Braille dots or pairs of Braille dots, and verbally reported the stimulus that best matched the reference stimulus. Answers were logged by the experimenter. Each of the 30 possible ordered pairs of non-identical stimulus were presented to the subject 3 times, for a total of 90 trials. The different pairs of stimuli were presented in randomized order.

3.3 Results

Figure 12 shows the results of the two tuning experiments. The preferred virtual dot width was found to be 2.0 mm, the most frequently selected parameter during the first step. The virtual intra-character separation was also found to be 2.0 mm, between the two most frequently selected values during the second step.

(a)	(b)
Figure 12: Results of tuning steps: frequency distribution of (a) dot widths, and (b) intra-character dot spacings.

These parameters corresponded to a intra-character dot spacing of 4 mm, compared to 2.29 mm for standard English Braille. The standard horizontal character-to-character distance of 6 mm was scaled accordingly to a virtual distance of 10.48 mm as illustrated by Figure 13.

Figure 13 (top): Illustration of dimensions of virtual Braille. Dot width is 2.0 mm. Distance between dot centers within a cell is 4.0 mm. Distance between the first dot of adjacent cells is 10.48 mm.

Figure 13 (bottom): Illustration of dimensions of standard Braille. Dot width is 1.5 mm. Distance between dot centers within a cell is 2.29 mm. Distance between the first dot of adjacent cells is 6.0 mm.

Figure 13: Dimensions of virtual and standard English Braille.

3.4 Discussion

The preferred virtual dot width of 2.0 mm turned out to be greater than the spatial period ε of 0.7 mm, leading to a strain pattern similar to that shown on the right-hand side of Figure 10. Moreover, since the dot width was more than twice the spatial period, the peak strain was lower than the maximum achievable. Assuming that loaded actuator deflections are half those measured without load (Section 2.1.1), this pattern resulted in strains in the order of ±20%.

Although the tuning experiments allowed us to find reasonable parameters, the tuning should ideally have been done either with a representative population of Braille readers or individually for each subject. Moreover, the use of texture on the dot may have contributed to an overestimation of the virtual dimensions. The two-hand comparison method may also have introduced errors, but constantly switching from a conventional Braille display to the VBD was impractical. Finally, the inter-character distance should have been tuned too. These coarse results were found to be sufficient given the scope and the aim of this study.