Optacon - Development of The Optacon

Development of The Optacon

With funding established, Bliss joined the Stanford faculty half-time, the other half being at SRI. At SRI tactile reading experiments were conducted to maximize the reading rates achievable with the Optacon, as well as development of the bimorph tactile array and the optics for the camera. At Stanford custom integrated circuits were developed including the silicon retina and the drivers for the bimorphs, since they required a higher voltage than normal for solid state circuits at that time.

The first technical challenge toward developing the reading machine was how to build a "tactile screen" that could create a dynamic tactile image which was perceivable by the user and that had a refresh rate fast enough for useful reading rates. Linvill's initial work with graduate students Alonzo and Hill indicated that a piezoelectric bimorph could be suitable as the transducer to convert an electrical signal into a mechanical motion. The advantages of bimorphs were efficient transduction of electrical to mechanical energy (important for battery operation), small size, fast response, and relatively low cost.

Alonzo determined that at vibration frequencies around 300 Hz, the amplitude needed for detection was much less than for frequencies around 60 Hz. Moreover, for reading rates of 100 words per minute, vibration rates of at least 200 Hz were needed. Linvill calculated the length, width, and thickness of a bimorph reed necessary for a resonance frequency of 200 Hz that could produce enough mechanical energy to stimulate a fingertip above the threshold of the sense of touch.

Based on these calculations, an array of bimorphs was constructed for reading rate tests with the computer simulation at SRI. The computer simulation presented tactile images of perfectly formed and aligned letters in a stream that moved across the bimorph array. Candy Linvill and other blind subjects learned to read text presented in this fashion with encouraging results. However, this simulation differed from the conditions that the user would encounter with an Optacon in the real world. There would be a wide range of type fonts and print qualities, plus the user would have to move the camera across the text rather than the computer moving the text across the tactile screen at a fixed rate. It wasn’t known how much the mental load of controlling the camera would reduce the reading rate.

In considering the transition from the text being presented by the computer to the user moving a camera across a printed page, Bliss realized that there was a critical flaw in the design of the Veteran Administration Stereotoner. Since English alphabetic characters can be adequately displayed with 12 vertical pixels, the Stereotoner designer had assumed that only 12 photocells would be needed in the camera. However, this assumes perfect alignment between the camera and the printed text, which is never the case with a hand held camera. When the alignment is random, as with a hand held camera, a well known engineering theorem states that twice as many pixels are needed. Therefore, the Optacon was designed with 24 vertical pixels instead of 12. This theorem isn’t applicable in the horizontal dimension, so the columns in a two dimensional array can be twice as far apart as the rows.

When a single column of 24 pixels is scanned across a line of text, all of the information is acquired. However, with the sense of touch, people are capable of perceiving two dimensional images. Bliss wondered if the reading rate would be higher if more than one column of 24 pixels were used, and if so, how many columns would be appropriate? Experiments with the computer simulation determined that reading rate increased dramatically up to 6 columns, which was a window width of about one letter space and this was about the maximum number of columns that could be placed on one finger. Jon Taenzer, one of Bliss’ Stanford graduate students, ran visual reading experiments on the same computer simulation and determined that for visual reading, reading rates continued to increase up to a window width of up to about 6 letter spaces. This led to a number of experiments toward trying to increase the tactile reading rate by increasing the number of columns in the tactile screen so more than one letter would be in view at a time. Instead of moving the text across only the index fingertip, tests were run with a screen wide enough for both the index finger and the middle finger to be used so two letters could be simultaneously tactually sensed. In another experiment the moving belt of text was run down the length of the fingers, rather than across them. The only approach that showed any promise of increasing the reading rate was when both index fingers were used, rather than the index finger and the adjacent middle finger. However, the use of both index fingers was incompatible with the design concept of using one hand to control the camera while the other hand sensed the tactile screen. The Optacon design was therefore based on an array of 24-by-6 pixels in both the camera retina and bimorph array.

Other questions had to do with the spacing between the tactile pins in the bimorph array and their frequency of vibration. It was well known from experiments reported in the literature that people could distinguish two points from one with their index finger when they were a millimeter apart. However these previous experiments had not been done with vibrating pins. What effect would the vibration have and was there an optimum vibration frequency? These questions were answered by experiments conducted by Charles Rogers, a Stanford graduate student working with Bliss.

While the neurophysiological data suggested that the smallest two point thresholds would be at vibration frequencies less than 60 Hertz, Roger’s experiments showed that the two point thresholds around 200 Hertz were actually smaller. Bliss hosted a conference at SRI, including some leading neurophysiologists and psychophysicists, to try to resolve this discrepancy, but no one had an explanation. From a practical standpoint, Roger’s result was very fortunate because the higher frequencies were required for refresh rates fast enough for reading up to 100 words per minute and for use of bimorphs small enough to construct a 24-by-6 array that fit on a fingertip.

The question of whether 144 tactile stimulators on a fingertip could be independently distinguished led to a confrontation at a scientific conference between Bliss and Frank Geldard, a University of Virginia professor. Geldard had written a major book on the human senses and was a leading researcher on using the sense of touch to communicate information. When asked how many tactile stimulators should be used in a tactile display, he maintained that no more than 8 tactile stimulators could be independently distinguished, and these should be on widely separated parts of the body. Bliss’ data showing useful reading with 144 stimulators on a fingertip appeared to be in conflict with Geldard’s research. The difference was between communicating using two dimensional tactile images versus an 8 point code. Both Bliss and Geldard were reporting similar reading rates, but in the days before high accuracy optical character recognition, the Optacon approach was much more practical.

These experiments determined the design parameters for Optacon’s man-machine interface: a 24-by-6 array of tactile stimulators, vibrating between 250 and 300 Hz, and with the rows spaced at 1 mm and the columns spaced at 2 mm (See Fig. 2).

In parallel with this human factors research was a pioneering effort to realize this design in a convenient portable unit, which would be critical for its success.