Everyone's effective mobility depends upon proper orientation; for main-stream society this is accomplished by printed signs. Print disabled, blind and visually impaired people are at a disadvantage for the lack of labels and signs. This paper describes an infrared system of remotely accessible signage. Infrared technology is compared with radio transmission, induction loops and computerized readers of passive signs. A key feature of the proposed infrared system is that signs are heard through a personal device, and that the user need only hear signs when he needs them.
The Rehabilitation Engineering Research Center at Smith-Kettlewell recognizes a need to outline the technical characteristics of infrared signage in order to promote proper understanding of its advantages and disadvantages. It is important to say at the outset that as an R&D facility, Smith-Kettlewell has an overriding concern that any implementation of a signage system not place constraints on future improvements, or preclude developments which could benefit our target population. As the place where these were first tried in 1979, the staff at Smith-Kettlewell feels qualified to address issues of their technical viability. Furthermore, the main engineer of the Love Electronics Inc., Mr. Julius Madey, worked at Smith-Kettlewell for a number of years, and we have made it a point to keep in close contact with him. Finally, various arrangements for specialized installations have been implemented by Smith-Kettlewell; our laboratory has experimented with the temporary installation of a tape-recorded infrared talking restaurant menu and infrared signage for hotel facilities at a recent national rehabilitation meeting Washington D.C. (RESNA). Most recently, we have collaborated with the San Francisco Department of Traffic Engineering in order to fabricate and install a full functioning infrared street/traffic-light signaling device at 5th and Market Streets in San Francisco. This system sends information to a hand-held speaking unit which indicates the street name, direction of travel and cross-walk status.
In 1978, William Loughborough, an engineer at our REC, experimented with beacons; the idea being to create target-practice and running games for visually impaired athletes. An infrared beacon and accompanying stereo headset was made with which a blind person could precisely align his head position to within a couple of degrees of the beacon.
About this same time, one of our blind staff members, diverted by sidewalk construction which involved adjacent streets, discovered himself to be lost at a time of night when no one was available to identify the street for him. Even though he is a good traveler with practiced techniques of regaining his orientation, he went through considerable trouble identifying his location.
It was then hypothesized that orientation and mobility (usually lumped together into one field) might be treated separately, as far as designing assistive devices is concerned. Several "mobility aids" had been developed which were primarily for obstacle detection; any aids classifiable as "orientation aids" were path-following aids--such as metal-detector canes intended for following paths marked with metallic strips--which were experimental in nature.
Having had this need so vividly demonstrated by that experience, engineers at Smith-Kettlewell surveyed "long-range navigation" systems and miniature gyro systems as possible components to a personal orientation device. Concurrently, engineers William Loughborough and Erich Sutter experimented with the infrared components used in the beacon as a way of transmitting sign information.
However, it was concluded that, even if a personal navigation system existed, much information would be unavailable to the user. The blind traveler has no direct access to signs, and signs carry a wealth of information--emergency exits, restroom locations, bus identification numbers, etc. It is unlikely that every kind of signage would ever be integrated into a navigation system, even if it were a full-featured implementation.
Experience with the infrared beacons led our engineers to the conclusion that infrared beams, coded with signage information, could be accessed from a distance, and could be localized using clarity of signal as the criterion. It was agreed that radio identifiers would lack precise placement information which infrared easily affords. Low frequency loops, a common system for information carriers in parks and historical sites, possess localization information, but cannot be heard from a distance (outside the loop).
On the face of it, one would think that radio-frequency (RF) transmitters would make poor signs, since their radiation might cover larger areas than infrared beams. Interference between signs would be greatly reduced if frequency modulation (FM) were used; any signal which saturates the limiter of an FM receiver "captures" weaker ones, and only the strongest one is heard. However, precise directional information about the location of the sign would not be possible unless a directional receiving antenna were used, and at radio frequencies, such an antenna would be large.
A radio-based signage system could be designed using receivers at the sign locations, assuming that only one transmitter -- the user's device -- were controlling them. These signs would be of the "transponder type," comprising a system that has its own disadvantages and which will be discussed later.
Some park exhibits use underground loops driven by amplifiers and tape players; the exhibit go-er carrying a receiver hears a recorded message whenever the pickup coil in his receiver is in close proximity to the loop.
This system has four disadvantages: First, the message carried by the loop cannot be heard from a distance, unless the loop were large enough to encircle that distance; if that were done, a specific target could not be localized. Second, the modern world is full of noisy alternating magnetic fields which would make noise in the user's receiver as he searches for "signs." Third, a loop is expensive to install; if not buried underground, it must be supported overhead in a horizontal plane. Fourth, several watts of power are needed to drive such a loop.
A working radio-based responder system was devised by Mr. Gary Kelly of Georgia Tech. In that system, the "signs" are receivers which await a code sent to them by a person carrying a transmitter. A touch-tone keyboard on the transmitter allows the user to enter a code for the object or location that he wishes; any "sign" (receiver) within range then triggers a beeper, or speaks through a loudspeaker, so that the interrogator can find the sound source.
The Georgia Tech system has two disadvantages: First, the user has to remember codes for signs he wishes to trigger. Second, the auditory outputs of the signs are heard by everyone. A simple transponder system would have severe limitations. Unless all signs in the immediate vicinity could be made to speak with the user's transmitter, an individual could not "browse" -- read signs at random to explore his environment. A more complicated transponder system could use a radio or infrared triggering system to activate an infrared sign which would be heard through a receiver in the user's device. This would be equivalent to the use of infrared signs as described in the Appendices A and B, except that both signs and hand-carried devices would be more complicated (see Sec. IV.A.4).
Obviously, an ideal device would be one which could read printed signs. Optical character recognition is not currently sophisticated enough to adapt to and correct for the parallax-induced distortions which result from variations in sensor/characters attitude.
A more likely future prospect is that special passive signs might be created. For example, "bar-codes" might someday serve as signs for print-handicapped people carrying portable bar-code readers. However, given the resolution of current bar-code reading heads, any bar-code installed today would have to be extremely large in order to be readable at a distance. As with computerized optical character recognition of printed signs, a bar-code system places the burden of precisely localizing the "sign" (before it is read) on the visually impaired user, and operation of the equipment would, for now, be temperamental.
Other approaches to passive signage will undoubtedly be up for serious consideration in the future. With the developments in electro-optics, digital signal processing, and learning networks, the mechanisms for detecting and interpreting information embedded in wavelength, polarization, phase, or holographic coding will follow.
We should have rules which encourage these developments for the future. We should not, however, wait for an ideal system to come into existence before we seek to implement currently available, adequate solutions.
First of all, the components of the Talking Signs, available from Love Electronics, are inexpensive and readily available. Installations are usually small; the main requirement is that the light-emitting diodes which generate the infrared signal be given a clear transmission path (they cannot be enclosed, or if they are, they must be behind a glass window; there is an installation in a San Francisco restaurant where the transmitter is attached to the inside of the front window by double-sided foam tape.)
The distance to be covered depends on the amount of power of the transmitter. The 5th and Market installation drives 24 LEDs with an average driver power of 4.8 watts. Indoor signs, such as door labels at Smith-Kettlewell, drive three LEDs with an average driver power of 450 milliWatts.
Within the determined range, the infrared "sign" can be detected whenever the receiver is within its radiation pattern. This radiation pattern is determined by the configuration of lights. For example, the street name, "Fifth Street," in the San Francisco installation is transmitted on a wide beam which can be received across the wide sidewalk of Market Street, and part way along the sidewalk of 5th Street. Status of the traffic light is transmitted over a very narrow beam which can only be received when the user is within the zone of the cross walk.
The sidewalk on Market Street is wide enough that a pedestrian can easily be out of the crosswalk zone if he were orienting himself by following along the side of the buildings. While outside the cross-walk, the user hears only the messages: "Fifth Street." (Pause.) "Zero-hundred block." (Pause.) "Facing east." (Pause.) When the user steps into the crosswalk zone, the pauses are filled with an announcement of the "color" of the light -- "red," "green," or "yellow."
Another possible installation of signs is one in which the traveler is "pointed" in the right direction to find, for example, the restrooms in an airport. For this, overhead signs with beams facing away from the restroom could be received as a person walks toward the restroom, but would not be receivable from behind. This is easily accomplished by pointing the emitters in the correct direction. A wide-angle view of such a sign in an open area would be arranged by "fanning" the beam out with more light-emitting diodes.
Although a responder system with public announcements using infrared technology could probably be implemented, the Love Electronics Smith-Kettlewell approach was to consider the orientation of a traveler to be a personal matter; we chose to present the sign information to the user through his pocket-sized receiver. Coincidentally, an amendment of the Americans with Disabilities Act, lobbied for by the National Federation of the Blind, guarantees a disabled person the right not to avail himself of an accommodation as he wishes. The implementation of a personal signage system complies with this provision in the law.
Clearly, a disadvantage of infrared signage is that obstacles, such as people, can be in the way. Yet this is not much different from obstructions that obscure visual signs. The user always has the option of holding the receiver overhead, if he suspects there might be an obstruction.
Technical implementation used by Love Electronice [with Talking Signs, Inc.] in their Talking Signs® product. There are advantages in retaining this implementation which are delineated below:
As stated in Appendix A, pulses of infrared are frequency modulated (FM). Early Smith-Kettlewell experiments determined that simple amplitude detection of the infrared beam was susceptible to 120Hz noise from room light, and the effect of sunlight was such as to prohibit the use of an amplitude-modulated system out of doors. The use of FM greatly reduced the effects of noise, and the exclusion of amplitude information below 18 kHz led to reduction of the effects of DC light current in the photodetector on front-end circuitry, thus permitting reception in the presence of bright light such as sunlight.
A well-known effect in the reception of FM is that an FM signal strong enough to cause amplitude "limiting" in the receiver will be the only one heard. This is called the "capture effect," and it means that there is a minimum of interference between transmitters.
If the user is in the presence of a high-powered sign bearing a street name, for example, but wishes to point the receiver toward a building entrance to hear the address (which would be emitted from a low-powered transmitter), the signal from the low-power sign will capture the receiver--becoming the stronger of the two, since the address sign would be more in line with the receiver's photodetector. When, by pointing the receiver in such a direction as to receive the address sign (at a level perhaps 50dB greater than the nearby high-powered street sign), the user will hear the message in his receiver change over to the address number of the lower-power transmitter.
Infrared was chosen because of the ready availability of emitters and photodetectors. It is well known that light-emitting diodes (LEDs) are most efficient in the infrared region, due to quantum physics of charge-carrier recovery. While visible light could be used, the selection of devices with high efficiency were chosen instead. Thus far, these generate light in the 950 nanometer wavelength. New devices are now coming out which generate light in the 880 nanometer region with even higher efficiency.
There are very sensitive photodetectors which work best at shorter wavelengths (in the region of visible light), but these tend to be slow in response, and detection of an FM carrier would not take advantage of their sensitivity. On the other hand, the same quantum physics that governs the efficiency of LED emitters also makes possible the creation of photo diodes which are at their most sensitive in the infrared region (below 900 nanometers in wavelength).
The technology of photodetectors and LEDs is largely driven by the telecommunications industry's use of fiber-optic lines; thus, matching sets of emitters and detectors are readily available. Fortunately for Talking Signs, large-area devices suitable for infrared communications through the air--not via fiber-optics--find application in TV remote control systems, cordless stereo earphones, and listening aids for the hearing impaired. Therefore, there are many devices to choose from to fabricate an infrared signage system, and available technology has made infrared the logical choice.
The possible selection of 880 nanometers (as opposed to 950 nanometers) will have no profound effect on reception, since photo diodes are rather broad-band in nature. However, the incorporation of narrow-band interference filters in the receiver front end would permit multiple optical carriers to be used.
A serendipitous advantage in choosing infrared is that light beams in this region are not visible. This greatly reduces the chance of vandalism, since the emitters do not draw attention to themselves. Needless to say, the absence of distracting visible light sources in the environment makes an infrared signage system surreptitious in nature.
Because large-area photodetectors stand to capture the most infrared light, they are the logical choice when fabricating receivers with optimal sensitivity. Unfortunately, the larger the area of the detector, the more junction capacitance adversely affects the response at high frequencies. Therefore, good engineering suggests the compromise of selecting a low carrier frequency, making this just high enough to avoid the audio spectrum of moderate-fidelity sound.
Twenty-five kiloHertz was chosen as the frequency we used to modulate the infrared beam. In the future, detector configurations will improve; higher sensitivity and multiple device configuration will mitigate the need for using large-area junctions. Detector frequency response is also expected to increase. However, adhering to 25kHz as a standard will have no adverse effect on new photodetectors as they become available.
Modulating the infrared beam at 25kHz avoids frequencies of infrared devices in current use--for example, the 40kHz emissions of remote-control units and FM carriers of devices for the hearing impaired. Moreover, the current choice of narrow-band FM (2.5kHz deviation) leaves open the option of adding higher-frequency subcarriers onto future signage systems.
In other words, infrared signage systems could be expanded to include dual-frequency or multifrequency receivers (reserving one frequency for emergency information, etc.). In fact, multipurpose receivers which could be used as cordless earphones and listening aids for the hearing impaired, as well as being Talking Sign receivers, are made possible by the current choice of carrier frequency and deviation restrictions.
An early decision was made to transmit voice information directly. An alternative would have been to transmit digital information that would require decoding in the receiver. Modulation of the transmitted signal by a voice recording was chosen to simplify the receiver design; a consideration that will become less important with large-scale integration of receiver components and computer memory.
It is acknowledged that direct transmission of the voice recording takes more power than would be required by transmission of a digital code. However, the power requirements are not now excessive. For example, a Talking Sign installed in a restaurant in San Francisco with sufficient power to reach beyond the curb, and with sufficient sidelobes to cover the distance to the adjacent shop doors, draws 30 milliamps. This includes the circuitry for "playback" of the recorded voice. (As a point of comparison, this device would run for 17 hours on an alkaline 9-volt battery.)
Initially, Talking Signs will be suitable only in locations where power is available. Their contribution to the energy load in powered installations will not be significant, considering the addition of 450 milliwatts, or even 5 watts (for a very high-power sign); this is in strong contrast to the load presented by public lighting (which may be thousands of watts). Smith-Kettlewell is now putting together a photovoltaic system based upon a transponder protocol which we believe may satisfy those situations where conventional electrical power is not available.
It is arguable that, in future installations, transmitting digital information may be an advantage. Besides the possible advantage in saving power, addressing receivers with alternative outputs--for the deaf-blind or multilingual speech synthesizers--may be attractive reasons for transmitting digital codes to so-called "smart receivers." Nevertheless, new receivers can easily be made so as to decode the original 25kHz FM transmissions; old signs need never be replaced.
Experience has shown that the transmitters are long-lasting. Early equipment at Smith-Kettlewell was kept running for at least eight years without noticeable degradation of performance. In contrast, receivers will be subject to wear-and-tear as they are carried by the users, and they are much more likely to need replacement than are transmitters. This is fortunate, since in a full signage system, there should be many more signs than print-handicapped people to "read" them.
As multichannel or multifunction receivers become available, it is only necessary to assure that these are compatible with early sign (transmitter) installations; this is technically easy to do, and it would assure that equipment of current standards will still be useful for decades to come.
We have argued in favor of the choice of infrared modulated by a 25kHz carrier which, in turn, is frequency modulated. In order for current products to address a wide range of signage needs, transmitters should have a design which afford the following features:
Both high and low power must be available.
To serve as directories, informational messages (such as parks and zoos) and to present restaurant menus, a means by which transmitters can be modulated by other audio sources must be specified. The digital audio storage in the transmitters should accommodate more than one message -- i.e., "going up," "going down," or "red," "yellow," and "green".
Recorded messages must be determined by the purchasers and if community response indicates that a message is wrong or confusing, it must be changeable.
The current transmitter product of Love Electronics implements those requirements as follows: An on-board LED driver transistor is sufficient to provide 45mA of average LED current; a range of from one to four LEDs covers most short-range signage applications. An out-board LED driver can provide an average LED current of 100mA; the inclusion of a separate power supply makes possible it possible to drive an array of 24 LEDs or more. Such a system can be used to cover a wide dispersion at medium range or can be faced in one direction to serve as long-range beacons (see sec. III.5).
The Love Electronics subcarrier oscillator can be separated from the on-board digital audio memory and fed from another source such as a tape player. For example, a sign in a restaurant could direct the visually impaired diner to an instrument which has a push button. When pressed, a tape mechanism would "read" details of the menu over the same infrared beam, heard by the user through his hand-held unit.
The digital audio memory on the Love Electronics board can be divided into as many as eight message blocks. A total time of 16 seconds is available; For example, the designer can have two messages of two seconds, one of four seconds and one of eight seconds on the same transmitter. These are selectable by electronic logic by way of an electrical connector on the board.
On the current product, messages are input from a microphone. The messages desired are specified and recording is done at Love Electronics. Smith-Kettlewell, on the other hand is building a speech recorder to load the digital audio memory, and the construction documentation for this speech recorder will be in the public domain. The erasable programmable read only memory (EPROM) which stores the speech can be unplugged, re-recorded and replaced.
An infrared system stands the best chance of any current technology to serve in implementing a signage system that is remotely accessible. Configuration of the transmitter beam, done by positioning of LEDs and directed by shrouds or lenses, gives infrared a clear advantages over radio transmission (directional antennas being comparatively large). The directionality of the photodiode in the receiver constitutes an inherent directionality with which signs can be localized. Storing recorded speech in memory provides ultimate flexibility as to what the signs can "say". Finally, these signs are private; The user looks for signs when he needs them and hears them through a personal device.
To date, three installations exist. The National Rehabilitation Hospital in Washington D.C. has some rooms and facilities labeled with Love Electronics "Talking Signs". The owner of a San Francisco restaurant considered it economically feasible to label his business with a Talking Sign. The San Francisco Department of Traffic Engineering has installed street name and traffic-light indicators at a major location in downtown San Francisco.
Finally, it is the opinion of the staff at Smith-Kettlewell that decisions made by the Love Electronics Inc. are not limiting to improvements in the future. The simplicity of current transmitters assures that future receivers can always be tuned to hear them, even if other carrier frequencies and modes are implemented.
William A. Gerrey
Rehabilitation Engineer