The Effect of Receiver Beamwidth on the Detection Time of a Message From Talking Signs, an Auditory Orientation Aid for the Blind.

Bo N. Schenkman
International Journal of Rehabilitation Research 1986, vol. 9(3), 239-246.
Copyright: Heidelberger Verlagsanstalt and Druckerei GmbH Edition Schindele, Heidelberg

Dr. B. N. Schenkman is at the Department of Psychology, University of Uppsala, Uppsala, Sweden.

ABSTRACT

Talking Signs, an orientation aid for blind people, delivering an auditory message through a loudspeaker in a receiver, presupposes that the user directs his/her receiver to a transmitter in order to get the message. The time of making this orientation of the receiver and detecting the message was studied, when the beam width of the receiver was varied. The results indicated that detection time decreased with increasing beam width, but the difference between the two widest beam widths was not significant. When considering also direction accuracy, it is likely that optimal detection time is gained with a beam width of about 24 degrees. It was also found, that some subjects decreased their detection time with training. No differences were found between blind and sighted participants.

1. Introduction:

A visually handicapped man or woman is at a particular disadvantage in the cultural environment of the modern human society. A large part of information relevant for the pedestrian's position is given in the form of visual signs with various coded information. Examples of such signs are street names in intersections or room numbers in a hotel. These problems of orientation for the blind are not solved by ordinary orientation and mobility aids for the blind such as long canes, nor by aids in some kind of map, either in auditory or factual form. These two groups of aids are reviewed by Farmer (1980) and Bentzen (1980), respectively.

An orienting device for blind people called Talking Signs, presently under development (Loughborough, 1979), will provide information for the blind user on his or her position in space. It is a system composed of two components, a transmitter and a receiver. The transmitter is constantly emitting infrared light. The blind person is searching for this light by turning the receiver towards the transmitter. The system makes it possible for the blind user to hear recorded messages via the receiver, which he holds in his hand. The audible message is stored in the transmitter on a solid state , "read only" memory. The audible message is modulated onto an infrared light beam of 950 nm at 25 kHz and emitted from a light-emitting diode (LED). The handheld receiver demodulates the stored message from the light beam and the user can thus hear, through the speaker in the receiver, the information that was stored on the solid state memory, eg. "floor number four". Both the transmitter and the receiver are in the sizes of 0.083 x 0.054 x 0.028 m.

Transmitters are planned to be permanently located at certain places of interest and the blind person to carry the receiver with himself/herself. By moving the receiver he or she may find the light beam from the transmitter and will thus hear the recorded message. The detection is the result of a search procedure, where the beam width of both transmitter and receiver will be of importance.

A wide angle of the receiver would make it easier for the blind to hear the message, as the detection time would decrease, but at the same time he or she would perceive the location of the transmitter with lower precision. This could be a problem for the blind user, if he or she wanted to move toward the place where the transmitter is located. Such a situation could exist e.g.. in a hotel, if the transmitter emitted the number of a certain room, and the user wanted to enter that room. An illustration of this problem is given in Fig. 1.

Fig. 1. Example of how a Talking Signs receiver with wide beam width offers less directional accuracy than a narrow band width when trying to locate the transmitter. A. Receiver with wide beamwidth. B. Receiver with narrow beamwidth.

2. Problem:

The ease with which a blind person would find the infrared light beam from the transmitter of the Talking Signs depends on various factors. One important factor is the width of the angle of the hand-held receiver, that accepts the light beam. it is expected that the larger the angle, tile easier it will be for the user, as measured e.g.. in detection time, to find the light beam. However, a wide angle would give less directional information on the location of the transmitter than a narrower angle, which is important in those cases, when the device is to be approached .

The main purpose of this study was to investigate the effect on the user performance of different angles of receiver beam width and also to determine the size of this effect for the different beam widths. The effect of training when using the Talking Signs system was also studied, as well as kind of subjects, viz. blind and sighted. Blind persons may perform differently with the system, as blind persons have been alleged to differ from sighted persons in how they orient in space (Worchel, 1951). Their spatial concepts might differ from those of sighted persons.

Dependent variable was the detection time, i. e. the time needed for the subject to find the infrared light beam.

3. Method:
3.1 Subjects:

There were two groups of subjects. One group consisted of six normally sighted persons, four men and two women, aged 16-37 years (Mdn - 27 years) and one group consisted of six blind persons, three men and three women, aged 28-39 years (Mdn - 31.5 years). One blind subject had some very minor light perception, but it was not necessary to blindfold her. All the blind subjects had been blind since childhood.

3.2 Measurement of the beam width of the receivers:

Three angles were used, and were designated "wide", "normal" and "isle':. The apertures of the normal and isle lenses were 25 mm. The aperture of the wide angle lens measured 5 mm. For optical reasons the apertures of the three receivers could not be made equal.

A LED, placed on a bench. was driven by a transmitter. The focal point of each lens on the receiver was placed 1.23 m from the diode. The beam width given by each lens was determined by moving the receiver around its center, until the signal from the transmitter was no longer detected. The beam width for the wide, normal and isle lenses was determined to be 46x, 23.5x and 9x, respectively.

3.3 Experimental arrangement:

The experiment was conducted inside a semi-anechoic chamber, its inner dimensions at the base being 16 ft. 6 in. x 7 ft. 8 in. (5.03 x 2.34 m) with a height of 8 ft. (2.44 m).

To make the test participants move in a defined space, a rope of small diameter, placed on the floor in the center of the chamber, formed a circle with a diameter of 23 in. (0.58 m). With their feet the participants could feel when they were about to step outside the circle. LEDs were placed outside this circle at a distance of 0.90 m from its center and at a height of 7 ft. (2.13 m). The diodes were made to point down at an angle of ca 30x. Each diode was attached to a cable that could be connected to a Talking Signs transmitter. The transmitter was next to the experimenter who sat 9 ft. (2.7 m) away from the subject. When attached to one of the four cables. the diode emitted the message stored in the transmitter. i. e. 'floor number four . The message had a duration of about 1 see. and was continuous as long as the receiver detected the infrared light beam from the diode.

3.4 Procedure:

The subject was first briefed on the nature and purpose of the experiment. He or she then entered the circle made by the rope and. if sighted. put on a blindfold. Before the first session the subject had a few practice trials. but no other practice was given before any other session. A richer description of the setup was given to :he blind than to the sighted participants. to allow both groups to obtain roughly the same amount of knowledge on the physical characteristics of the experiment.

The task given to each individual required the detection of the emitted signal by moving around within the circle. Raising the receiver above the head was not permitted. The time interval from the command to start the search to the detection of the complete auditory message. i. e to the word four, was the dependent variable. Technical defects in the transmitter made it impossible to have a clear message all the time: a very distinct and easily recognizable noise secured in about 10 Car of the trials. When this happened the clock was stopped as if an auditory message had been emitted. These trials were included in the data. as there was a clear break between silence and noise. At each session the subject cot one of the six possible orders of the three beam widths. An experimental session consisted led of 24 trials with one beam width. a rest interval. 24 trials with another beam width, a rest interval and 24 trials with the third beam width. At each trial. only one of the four diodes was connected. but their order of presentation was randomized. Two new sessions. designed in a similar way but with new randomization. were conducted for the sighted participants. The blind subjects had only one trial-session.

4. Results and Discussion:

The mean detection times for the 24 trials were calculated for each beam width, session and observer. The means for the sighted and blind subjects are presented in Table 1. The statistical analyses were done both separately and for the two groups combined.
Table 1. Mean detection times (sec) for the subjects given three receiver detector angles: 9 deg, 23.5 deg and 46 deg. Blind had one session and sighted had three. Table is a matrix of means (M) and standard deviations (SD) for each angle and each session.

				M	SD	M	SD	M	SD	M	SD 



Blind session			25.2	19.9	18.4	10.5	12.1	13.2	18.7	14.0



Sighted session 1		20.2	13.1	11.3	7.4	7.7	3.2	13.0	5.1



Sighted session 2		12.2	6.0	10.3	4.9	7.5	5.5	10.0	5.3 



Sighted session 3		13.5	7.6	8.3	3.5	6.8	3.0	9.5	4.3




Analysis of variance, randomized block factorial design and mixed model was performed for the sighted observers. The block factor, i. e. the subjects, was the random variable. Table 2 indicates that there were significant differences between the beam widths at p.05. T-tests showed there was a significant difference between the wide and narrow beam width means, t(l0)=5.11, p.05, and between the narrow and middle beam width means t(l0)=3.38, p.05. The performance of the sighted participants with the three beam widths at the three sessions are presented in Fig. 2.
Figure 2. Mean detection times for the sighted participants at each of the three sessions and for the blind participants at their single session with the three beamwidths of the receiver.

Table 2. Analysis of variance, mixed model, on the mean detection times for the sighted subjects. Table is of three parameters: df, MS and F for seven sources of comparison.

Source			df	MS		F



Subjects (Su)		5	226.99		12.31*



Session (Se)		2	66.49		0.92



Angle (A)		2	295.65		13.39*



Su x Se     		10	71.95		3.90*

	

Su x A     		10	22.08		1.20



Se x A			4	29.36		1.59



Su x Se x A		20	18.44		___


*p0.05

The only significant interaction was the one of subjects with sessions, p.05. This indicates that the subjects benefited differently from practice. In the model chosen performance over the three sessions was not significantly different. The mean detection times for the blind observers at the three beam widths are shown in Table 1 and Fig. 2. The means were analyzed by analysis of variance, which is summarized in Table 3. The performance of the blind participants, which fell into two distinct groups, showed a significant difference at p.05. The effects of the three beam width angles were also significant, p.05. Testing for differences between the means of the angles showed only that between the wide and narrow angle to be significant, t(10) = 2.92, p.05.
Table 3. Analysis of variance, mixed model, on the mean detection times for the blind subjects. Table is of three parameters: df, MS and F for three sources of comparison.

Source			df	MS		F



Subjects (S)		5	703.91		12.63*



Angles (A)		2	238.06		4.27*



S x A			10	55.72		___


*p0.05

The mean detection times for the blind subjects were analyzed together with the values of the sighted subjects in the first session of the earlier experiment according to a split-plot factorial design. The analysis is presented in Table 4. The type of subjects, i.e.. blind or sighted and the angles were seen as fixed factors, while the subjects were a random factor. The algorithm used was the one described by Kirk (1968, 248).
Table 4. Analysis of variance, mixed model, on the mean detection times for sighted subjects in the first session and the blind subjects. Table is of three parameters: df, MS and F for seven sources of comparison.

Source			df	MS		F



Between subjects	11	___		___



Type of subjects: 



  blind or sighted (T)	1	289.57		0.64



  Subj. w. groups	10	454.92		___



  Within subjects	24	___		___



Angles (A)		2	480.03		10.09*



T x A			2	4.5		0.09



A x Subj. w. groups	20	47.59		___


*p0.05

The blind subjects fell into two sub-groups, one group of three that had short detection times overall and one group of three having long detection times. A simple explanation for this could not be found. As can be seen from Fig. 2: the blind needed 5-7 see more than the other group in session 1, to detect the light beam and get the message, but the differences between the two groups of subjects were not found to be significant. The form of the cues for the two groups in Fig. 2 is very similar.

The present setup had the diodes pointing down at the same angle, 30x, toward the subject, and this probably facilitated his/her task. The detection times would have been longer with varying angles of these diodes.

5. Conclusions:

The results demonstrate, as expected, that the wider the beam width of the receiver, the easier it will be to find the light beam from the transmitter. It should however be observed that increasing the beam width beyond 23.5x did not improve the performance of most participants as markedly as between 9x and 23.5x (see Fig. 2). The steepest reduction in detection time was between the 23.5x and 9x beam width. A narrow angle probably makes the task more difficult and would frustrate the person in a real life situation. An angle around 24x of the receiver is therefore probably the best choice, when weighing the factors of time of detection and ease of localization of the Talking Signs. A more narrow angle would make it easier to know the location of the transmitter, but this advantage would be mitigated by the longer detection times. This conclusion is tentative, as no data were collected in this study on the accuracy of localization.

No significant differences were found between the blind and the sighted subjects. This is analogous to the results of Fisher (I964) who compared spatial localization of blind and sighted subjects, but did not find the blind to localize better than the sighted either auditorily or tactile-kinesthetically. The interaction of training with subjects was significant, which indicates that the training had an effect on some, but not all, of the subjects.

How the Talking Signs will function in real-life situations remains to be seen. Other variables such as optimal range, speech intelligibility etc. need to be studied before its final implementation.

6. References:

Bentzen, B. L. Orientation aids. In Welsh, R. L. & Blasch B. B. (Eds.), Foundations of orientation and mobility. New York: American Foundation for the Blind 1980, 291-355.

Farmer, L. W. Mobility devices. In Welsh, R.L. & Blasch, B.B. (Eds.), Foundations of orientation and mobility. New York: American Foundation for the Blind, I980, 357, 412.

Fisher, Y. H. Spatial localization by the blind. American Journal of Psychology, 1964, 77, 2-14.

Kirk, R. E. Experimental design: procedures for the behavioral sciences. Belmont, CA: Brooks-Cole, 1968.

Loughborough, W. Talking Lights. Journal of Visual Impairment and Blindness, 1979 73, 243.

Worchel, P. Space perception and orientation in the blind. Psychological Monographs, 1951, 65, 1-28.

Acknowledgments:

This study was partially funded by a Thord-Gray fellowship from the American-Scandinavian Foundation and partially by a grant from the National Institute of Handicapped Research. The experiments were made possible by the generous cooperation of the staff of the Smith-Kettlewell Institute of Visual Sciences in San Francisco: the late Mr. Gordon Holmlund, Dr. John Brabyn, Dr. Deborah Gilden, Mr. William Loughborough and Dr. Brian Brown. Dr. Gunnar Jansson and Mr. Kurt Vikman from the University of Uppsala, Sweden, gave many helpful comments.

Author's address:

Dr. B. N. Sehenkman, Department of Psychology, University of Uppsala, P. O. Box 227 S-75104, Uppsala, Sweden.