Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Chapter 6 THE TECHNOLOGY OF ELECTRONIC TRAVEL AIDS On the basis of some assumptions regarding the types of environmental information needed to enhance independent travel by visually impaired people, this chapter provides a review of mobility aid technology to date. We discuss strengths and weaknesses in the information acquisition and display of these devices, and we make suggestions for research aimed at improving on past aids, along with speculation regarding existing and future technology that could be applied to the problem. INFORMATION NEEDS OF THE: PEDESTRIAN As previous chapters have discussed, it is generally accepted that, in order to engage in safe and efficient travel, the pedestrian must have access to certain categories of environmental information. Sug- gestions regarding the information needed have been put forward by Foulke (1971), Kay {1974), Strelow (1985), and others and usually include the following concepts: (1) Information is needed regarding the presence, location, and preferably the nature of obstacles immediately ahead of the traveler, from ground level to head height and over a wide enough area hori- zonta]ly to cover the width of the traveler's body. The minimum distance or range over which this information is needed is a com- fortable stopping distance at normal walking speed. A greater range is desirable. (2) Information regarding the path or surface on which the traveler is walking is highly desirable; this includes texture, gradient, and upcoming steps (both up and down) and boundaries to left and right (including step-downs at sidewalk edges). (3) Information regarding the position and nature of objects to the sides of the travel path is desirable. This includes hedges, fences, doorways, trees, etc., forming part of the shoreline on either side of the path. (4) Other information to enable the traveler to maintain a straight course is extremely helpful, notably the presence of some type of aiming point in the distance, often provided in practice by distant traffic 67
68 sounds. Knowledge of absolute or relative direction of travel is also helpful. (5) Information on landmark location and identification is needed. This can include information under the above categories (especially 3) and also includes the ability to positively identify specific environmental features such as building entrances, room numbers, elevators, rest rooms, floor numbers, intersections, etc. (6) Sufficient information must be provided by one means or another to allow the traveler to build up a mental map, image, or schema for the chosen route to be followed, including turns and other discontinuities. PAST AND PRESENT MOBILITY AIDS With the above assumptions in mind, we examine the progress to date in technology designed to acquire and display at least some of the above types of information. The Long Cane The long cane, introduced in its present form by Richard Hoover (1950) in the late 1940s, effectively allows detection of obstacles within a 3-foot range and tends to warn other pedestrians to get out of the way. The former feature requires fast stopping reaction or slower walking speed when an obstacle is encountered, while the latter does not add to the grace of independent travel. The skilled user of a cane can, however, acquire directly most of the information needed regarding obstacles and path or surface. In addition, location cues from the sounds emitted when the cane tip (especially a metallic tip) contacts the ground can provide a surprising amount of the additional information regarding objects at the sides of the path. These subtle cues are frequently unknown to, or underrated by, the sighted community (except for mobility specialists) and their presence provides a powerful incentive for skilled blind pedestrians to eschew electronic travel aids. These subtle cues and their use receive a great deal of attention in orientation and mobility training. Electronic Travel Aids Following the development of radar and sonar technologies for remote sensing during World War II and the introduction of transistor technology, which made portable electronic devices practical, inventors began to see the potential for various obstacle-detection devices to aid blind people. A huge number of such devices were developed, mostly using the transmission of an energy wave (usually ultrasonic) and the reception of echoes from objects in or near the traveler's path. The choice of ultrasound for the transmission medium, as opposed to light or radio waves, was dictated by the convenience with which echo ranging
69 could be performed in this medium due to the relatively slow speed of propagation. Optical sensing was used in some cases, when the operation of the system was passive or when range measurement was not a design goal (e.g. in the Laser Cane). Sonar technology, as it existed by 1960, was well developed for underwater purposes, but its use above water was complicated by much- increased signal attenuation and the difficulty of coupling a transducer effectively to the air. These problems were overcome, and ultrasound became the most popular sensing medium for mobility aids. The main limitations of the technology were limited useful range and difficulties experienced with reflections from smooth surfaces (which act like mirrors to ultrasound due to its long wavelength). Clear-Path Indicators Versus Environmental Sensors Once the technical problems had been solved, the principal arguments among developers revolved around the amount and type of information that is desirable to present to the user and the manner in which it should be presented. Development followed two schools of thought. One class of aids, known as obstacle detectors or clear-path indicators (Foulke, 1971), warned only of the presence and sometimes the approxi- mate range of obstacles directly in the travel path, while generally not being concerned with identification of the obstacles detected. Such devices generally had the advantage of relatively lower cost. Included in this category is the Russell Pathsounder (Russell 1965~, which uses a 30-degree ultrasonic beam transmitted from a chest-level unit suspended from the user's neck. It can provide tactile and auditory warning of objects up to 6 feet ahead and allows some range estimation. The Mowat Sensor (Presley 1977) is a hand-held, pulsed sonar system with tactile output. The narrow-beam, flashlight-sized device vibrates with a period proportional to target range. The Nottingham Obstacle Detector (Armstrong 1973) is similar in operation and shape to the Mowat Sensor, except that it provides an auditory readout of range with eight notes corresponding to the musical scale. The Laser Cane (Nye, 1973) is a conventionally shaped long cane with laser emitters and receivers aimed to detect overhangs, down curbs, and targets straight ahead within a selectable range of 6 to 12 feet, giving auditory and tactile warnings. A second category of aids, known as environmental sensors (Foulke, 1971) attempts more than mere detection of obstacles. The first of these to be developed--and one of the most well known--is the Kay Sonic Torch (Ray 19641. One of the few devices designed to replace rather than merely supplement the long cane or dog guide, the Sonic Torch {which is no longer in production) was a hand-held, narrow-beam ultrasonic device with an auditory output presented via an earphone. The Sonic Torch, in contrast to those ultrasonic travel aids of the obstacle-detector class (which normally use simple single-frequency pulse transmission), transmitted a wide-bandwidth (40-80 kHz), frequency-modulated, ultrasonic energy wave. Reflected signals from the Sonic Torch were converted to the audible region by multiplication
70 with the transmitted signal. The result was an auditory signal extremely rich In information--the pitch corresponding to range and the timbre corresponding to variations in target surface texture, enabling target identification. Later developments in this concept led to the Binaural Sensory Aid (Kay 1966), known commercially as the Sonicguide. Employing the same methods of transmission and reception as the Sonic Torch, this device looks like a pair of spectacle frames; it is worn on the head with a wide-beam (60-90 degree) transmitter. The two receiving transducers are splayed a few degrees to the left and right, and their signals are presented separately to the left and right ears, giving a built-in direction cue. The Sonicguide is used in conjunction with the long cane or dog guide, like most other electronic mobility aids. Recent Developments The first generation of mobility aids, described above, was criticized on grounds that included cost-effectiveness, the masking of natural echo-location cues, and the necessity of training for the more complex aids. In the late 1970s and early 1980s, device developers came up with some new solutions and refinements to address these problems. Criticism of the degree of information presented by the Kay Binaural Sensory Aid was answered by the development of a prototype single-channel system with a narrow beam width, eliminating the constant signals returned by objects in the periphery, while retaining the advantages of the binaural version for object identification. To shed light on the controversy regarding the optimal form of range cue for an obstacle detector, the American Foundation for the Blind developed a microprocessor-assisted ultrasonic ranging device (Maure et al., 1979) using the sonar electronics from the Polaroid auto-focusing mechanism. The microprocessor can be programmed to present its output in any one of a variety of auditory codes, including the spoken voice. Other Polaroid-based ranging systems have also been developed (Heyes, 1982~. In the Federal Republic of Germany, the Siemens Company produced its own spectacle-worn and hand-held obstacle-detector systems, while in England the developers of the Nottingham Obstacle Detector devised a modified head-worn version, the Sonic Pathfinder {Heyes, 1984), designed to allow discrimination of target direction while remaining basically an obstacle-detection system. Investigations were made of the feasibility of tactile information presentation in a mobility aid, in order to overcome the criticized masking effects of auditory displays {and also for use by people who are both deaf and blind). Early research in this field at Smith- Kettlewell and elsewhere (Collins, 1967; Collins and Madey, 1974), centered on the direct presentation of television camera images on the skin, on a point-for-point basis using large (32 x 32 or 20 x 20 point) tactile arrays on the abdomen. As discussed in Chapter 5, it was concluded that this concept has educational value when confined to
71 simple images and shapes, but the complex scenes encountered outdoors cannot be readily interpreted through the skin for use in travel. Two alternative approaches were then investigated for preprocessing the tactile information, thus simplifying the display. The first approach used a scanning sonar system (Brabyn et al., 1981) to present a plan-type view of the environment on the skin, displaying range and azimuth information in the same manner as a PPI radar display. This approach was successfully tested, but the expense of the large tactile arrays militated against successful commercialization. An alternative approach, using computer processing of video images to extract only the features vital to mobility (such as curbs, poles, fences, etc.), was then tested. Range and direction were coded on a one-dimensional tactile array, and supplemental synthetic speech information was used in the system output {Collins and Deering, 19841. In order to overcome difficulties experienced with ultrasonic obstacle sensors when attempting to detect smooth surfaces at oblique angles, an infrared ranging system was designed at Smith-Kettlewell. The prototype device uses a disparity technique that does not require complex electronics for propagation delay measurements. The initial version uses an auditory output, with an inverse relationship between pitch and range. The beam is extremely narrow (approximately 5 degrees), and smooth surfaces such as linoleum can be detected at angles up to approximately 60 degrees from the perpendicular. Orientation Aids Orientation involves two aspects: (1) area familiarization--becoming acquainted with a geographical area or with routes within it and (2) navigation--finding one's way from location A to location B. Navigation can be accomplished either by keeping track of one's position with respect to prominent landmarks or by following a series of rote directions. The primary aids used in area familiarization in the past have been the use of a sighted guide or a factual map. Experience and research have shown factual maps have met with only limited success in familiar- izing blind people with a complex environment, such as a city or college campus. Several electronic aids have recently been developed to assist in both area familiarization and navigation. The "Talking Signs" system (Loughborough, 1979), developed at Smith-Kettlewell, uses a network of low-cost infrared transmitter modules placed on normal navigational signs in the environment, such as street corners, bus stops, corridors, and room and building numbers. A voice output corresponding to the wording of the sign is produced by a a hand-held receiver, which receives the infrared transmissions and converts them into a spoken message. Each transmitter contains a computer memory chip on which its message is stored. This system is now in commercial production. A similar idea is being pursued by the Georgia Institute of Technology in the development of its Sonic Orientation Navigation Aid (SONA) system (Kelly, 1981~. This orientation aid uses radio-activated
72 auditory beacons to accomplish the task. If, for example, a user is searching for an elevator, he or she enters a corresponding numerical code on a key pad mounted on the hand-carried transmitter. The trans- mitter then interrogates any receivers in the vicinity. If a receiver is present on an elevator nearby, it will accept the code and produce an auditory beep. Obviously, successful dissemination of such orientation and navigation systems is problematic, due to installation costs. Even though substantial and expensive modifications to the environment have been made for the benefit of wheelchair users, it does not currently appear likely that the same types of changes will be made for the blind population. Optical Aids for Low Vision While low vision encompasses a wide range of visual impairment, existing optical orientation and mobility aids may nevertheless be divided into four types: distance magnifiers, minifiers, absorptive filters, and image intensifiers. In many cases an aid is designed to serve more general purposes for the individual than orientation or mobility. Distance magnifiers (telescopes) (Bailey, 1984) are used for the identification of environmental features such as signs, room numbers, etc., and in orientation to identify hedges, fences, doorways, trees, etc. Considerable skill is required in the use of these aids for the following reasons. First, the optical leverage of a telescope amplifies small hand tremors into large image translations, making their use suitable only for those with a steady hand (which older people are less likely to have). Second, because the image through a telescope portrays only a small area of the environmental field, it is difficult to aim the device at a desired object. To read a street sign, for example, the user may have to track along the length of the sign pole as an aid to sighting. Third, focusable telescopes are difficult and inefficient use, especially when the image of best focus is degraded by the very optical and neural impairments that bring about the need for a telescope. Additional limitations of telescopes are that: (1) considerable light loss lowers the image luminance, reducing visibility of details and overall effective contrast and (2) for fixed focus (to infinity) telescopes, minimum usable range is restricted to greater than 6 meters. A second class of optical aids for low vision includes image minifiers' such as reversed telescopes. While distance magnifiers expand a small portion of the field to a size that allows identification for those with central field and/or resolution losses, minification aids compress a wide field to a small central region of the retina. They are designed to allow detection and identification of objects on the periphery of the travel path for individuals with field constric- tions. Minification aids suffer from the same light loss problems as forward telescopes and have not been well accepted in practice. Users to
73 also find the unnatural optical flow pattern disturbing. The user is inadvertently retraining the vestibular-ocular response with these aids and may find that they produce dizziness and even nausea. A variant of this technique is the Amorphic Lens by Designs for Vision, which compresses the field only along the horizontal dimension, presumably on the theory that the view along the sides of the observer is more important for mobility than that in the vertical periphery. The Amorphic Lens produces somewhat disturbing distortions of the field, as the relative sizes of objects change with continuous rotation of the head, but it is considerably more graceful to use than spectacle-mounted or hand-held magnifiers or minifiers. A third class of aids is the absorptive lenses. It includes sunglass and Polaroid filters, which reduce ordinary glare from sunlight and specular reflections, and filters that selectively absorb light from one part of the visible and/or invisible spectrum. Selective filter are used primarily to remove short wavelength light that scatters more easily than long wavelength light and to remove ultraviolet light that with some pathologies can produce fluorescence of the media. Light scatter and fluorescence, of course, degrade the retinal image and reduce overall contrast, and such filters are designed to alleviate such effects. An important trade-off, however, is the problem that reducing the light with filters also reduces effective contrast of the image. The fourth class of mobility aids for low vision is the image intensifiers, intended especially for those with retinitis pigmentosa. These range from simple devices, such as wide-angle lamps {Morrisette and Goodrich, 1983), which increase illumination and hence effective contrast on and around the travel path, to more sophisticated devices, like the ITT Nightscope (Berson et al., 19731, which displays a video image of the system travel path sensed by a photo multiplier that is more sensitive than the eye. These aids suffer from a lack of grace and are effective only to the extent that the user can see and interpret the information acquired visually. LIMITATIONS OF EXISTING TECHNOLOGY As stated at the outset of this report, the functional utility and user acceptance of mobility aids should be considered in the context of the four criteria for successful mobility: safety, comfort, grace, and independence. None of the aids produced to date has achieved broad market penetration. In part this has been due to the resistance of some groups of consumers to use or support the use of electronic mobility aids, and in part to the limitations of the technology itself. There are a number of probable reasons for this phenomenon: · Information Limitations. It is evident that each of the approaches taken to date by designers of mobility aids is capable of acquisition and display of only a subset of the required mobility information outlined earlier. No device by itself--including optical aids or the long cane--provides all the required or desired information.
74 · Expense. All commercially available electronic travel aids are relatively expensive, costing between $300 and $3,000. · Relation to the Long Cane. Most are designed to supplement (rather than replace) the long cane or dog guide, and there is disagreement over whether the additional information they provide is worth the very considerable extra cost and the effort of training. A skilled user of the long cane can use the sounds emitted by the metallic cane tip for natural echo-location, providing a surprising amount of information about the immediate environment. This category of user, although not typical of the general population, would require substantial additional input from an electronic aid for its use to become worthwhile. · Auditory Cues. The auditory displays of many aids may mask the subtle echo-location cues mentioned above as well as important ambient sounds. · Training Difficulties. For the more complex aids, the training required to master their use generates resistance. This appears to apply to many optical aids as well as electronic ones. · Conspicuousness. Most aids to date for both blind and partially sighted people are conspicuous (not meeting the criterion of grace), and many are not conveniently removed or deactivated when not needed. · Limitations of Scale. Those aids developed to date addressing orientation problems would be very expensive to install on a large scale. · Lack of Performance Measures. There has been little success in demonstrating that optical or electronic aids for blind and partially sighted people actually improve their mobility. This lack of success reflects the crudeness of measures as well as the limitations of the aids themselves. Although these speculations are offered the reasons for the lack of acceptance of optical and electronic aids for mobility, not all the reasons are well understood--particularly in the area of magnifiers and minifiers for partially sighted people. In summary, the market need appears to be for mobility aid technology that is inexpensive, easy to learn, and not distracting from natural cues. The technology should, however, display significantly more information than what is available from natural cues and be able to give a demonstrable improvement in travel performance. RECOMMENDATIONS We can expect further improvements in electronic mobility aids to be limited less by technological considerations than by two other main factors: the determination and definition of the information needed by the traveler and the capacity of the nonvisual senses to process the information via a suitably encoded display. Compared with these problems, the development of technology to acquire and encode the information is relatively straightforward. Solution of the problems,
75 in turn, will depend on our ability to evaluate mobility performance and real needs for information {Leonard and Carpenter, 1973; Brabyn and Strelow, 1977; Dodd s et al., 19831. On the basis of this review, we make research recommendations in a number of areas. Problem Definition Information Needed for Mobility The critical types of environmental information needed for mobility (and their relative priorities) must be established by experiment, giving a more concrete starting point for the problem of information acquisition and display. A preliminary analysis suggests that this information should include obstacle detection and identification, path surface discontinuities, shoreline information and other n straight courses input, and landmark identification. RECOMMENDATION: Research should be undertaken to establish the types of environmental information needed to address the problem of information acquisition and display in the development of electronic mobility aids. Research on User Skills As a prerequisite for electronic aid design, research is needed to investigate specific aspects of existing user skills, including deter- mination of the prevalence of echo-location skills among the blind population and determination whether such skills would be significantly improved by use of metallic cane tips, improved training methods, or both. The importance of these subtle echo-location cues needs to be determined, since certain electronic aid displays can mask them. Research is also needed to determine the precise ways in which different low-vision pathologies affect mobility performance. RECOMMENDATION: Research should be undertaken on the relationship between user skills and electronic travel aid design and display characteristics. Technology for Display Design and Simulation RECOMMENDATION: Simulation of information displays should be developed. Different methods of displaying information for mobility must be carefully designed to match the sensory systems and tested by experiment (largely without developing sophisticated hardware). To achieve this, displays can be simulated manually or with simple technology, and their effects on mobility performance can be measured.
76 Technology for Optical Aids RECOMMENDATION: Research should be conducted to investigate why existing aids are rejected as well as the conditions required for acceptance. These results will be essential to the design and use of future travel aids. RECOMMENDATION: The visual requirements for mobility should be explored fully. Visual parameters to be studied include the necessary and sufficient sizes and locations of visual f ields and contrast thresholds for mobility and ocular motility, including the ability to scan the environment and track moving objects. Recommendation: Research is needed to study the kinds of image transformations that will be both useful and acceptable with minimal disturbance to the user: Research and development of innovative lens systems is needed. The Amorphic Lens, described earlier, might perhaps be improved on by using a nonuniform compression of the periphery that would give priority to the most important areas of the periphery for mobility. Research is also needed to determine how much image compression can be tolerated in a minifier, producing a wider field in which obstacles can be better detected without destroying the usefulness of the central field in identifying objects or tracking an aiming point. Research is needed to study the application of auto-focusing techniques, such as those found in cameras, to be used in future telescopic aids. Further research is needed to investigate both the objective assessment of glare and the measurement of image degradation as a function of wavelength composition of the light. These results would facilitate a rational choice of absorptive filters to produce the most effective attenuators of glare with the least amount of overall light loss. Technology for Information Acquisition Once the critical information required for mobility has been established in accordance with our recommendation pertaining to the identification of information needed for mobility, a wide variety of technologies should be explored for the acquisition of this information from the environment. Optical, Opto-electronic, and Other Electromagnetic Sensing Systems The use of miniaturized solid-state video cameras coupled with miniaturized image processors offers one flexible method for information acquisition. In addition to optical sensing systems, other sensing systems based on electromagnetic phenomena may also be
77 applicable. These include millimeter radar, infrared and heat sensors, and magnetic sensors. Light amplification systems can also be useful · . · · In . tow vlslon. RECOMMENDATION: Researchers applying electromagnetic sensors should be urged to find a way to process and display the information auditorially or tactually in an easily interpretable manner. - Sonic and Ultrasonic Systems Sound-based sensing systems suitable for information acquisition in this context may include use of both audible and ultrasonic sound waves. Appropriate audible signal transmission and reception may improve echo-location abilities. RECOMMENDATION: We recommend that research and development programs be initiated to adopt ultrasonic transmission and reception systems for both simple and complex degrees of information acquisition, ranging from simple pulsed transmission and reception to phased array, variable beam width, focused or unfocused scanning systems. Furthermore, we recommend that new developments in ultrasonic technology, currently in research use for spatial sensing by blind children, be carefully evaluated for possible application to mobility. New developments in wide-band frequency modulated ultrasonic spatial sensors (which have a built-in ability to display, auditorially, a relatively large quantity of information) include greater resolution, adjustable range, automatic level controls, and the simulation of the human retinal response by the provision of a narrow central beam and width (and weaker) peripheral beams that can be switched off or on at will. These new systems should be thoroughly investigated to determine their value in new mobility aids. Information Acquisition for Orientation and Navigation Aids Considerable research has already gone into the development of automatic navigation systems for military and civilian applications (e.g., the navigation of cruise missiles, ships, aircraft, and ground vehicles). The two essential components of an automatic navigation system are (1) a digital "map" representation of the geographical region of interest and (2) a functional module that senses the vehicle location via some combination of inertial, satellite, or radar sensors. The two components together might find use as a navigation system for the blind population. RECOMMENDATION: We recommend that exploration be made of the potential application of digital map and positive sensing systems to navigation and area familiarization for blind or visually impaired people. Information acquisition technologies that could be applicable
78 include inertial sensors, satellite and radio navigation data, footstep measurement and monitoring devices, and electronic directional devices, including compasses. Improved and less expensive methods of providing environmental labels, such as solar-powered coded transmitters or passive coded reflecting strips, which could be interrogated by intelligent transceivers carried by the pedestrian, should also be investigated. Technologies for Information Display Auditory Displays RECOMMENDATION: For auditory information display, the use of synthesized speech needs to be tested {this can be done by simulation, as mentioned earlier). Brief verbal messages, at a level of complexity selectable by the user (or in response to user interrogation) may provide a means of display for part of the information generated by acquisition systems. Speech is a very redundant means of communication. other auditory codes could well be more efficient {but may require more training). RECOMMENDATION: Efforts should continue to devise nonverbal displays having simple, easily learned underlying auditory codes but from which progressively greater degrees of information can be extracted as experience accumulates (such as that used in the swept- frequency ultrasonic family of devices). Simpler coding systems conveying less information may be useful in conjunction with other natural or artificial sensory inputs. RECOMMENDATION: Research and development projects are needed to investigate the use of externalized or stereophonic auditory displays. These displays have the potential of providing the user navigational information in real time both for environmental familiarization and for inclusion in new mobility devices. These investigations should study both auditory and synthetic speech displays. The goal of the research and development effort would be to make the sounds appear to be coming from different locations in auditory space. Tactual Displays Tactual displays can make use of both cutaneous and kinesthetic information and include tactile maps and simple vibration. Past experience shows that direct, unprocessed transfer of video images to large two-dimensional tactile array is not suitable for mobility. Furthermore, from a technical standpoint, it should be noted that large tactile arrays have been, to date, unreliable. However, new methods of coding or preprocessing spatial information before it is presented to such displays, and new display technology that makes two-dimensional arrays more technically realistic, should not be ruled out.
79 More modest use of tactile displays, such as in one-dimensional arrays, has proven technically feasible and capable of transferring considerable information, especially in the field of auditory substitution. This type of display should therefore be explored, along with single-point stimulation or variable-frequency vibration. Dynamic computer-controlled factual maps, should a feasible technology emerge for their implementation, could be useful in orientation training and research. RECOMMENDATION: We recommend that research be continued on new methods of factual display, emphasizing preprocessing and ease of interpretation of displayed information. Research and development efforts should be initiated to investigate the feasibility of new tactile display technology with the goal of fabricating dynamic computer-controlled factual maps to be used in mobility training and research. Visual Displays Even the most rudimentary residual visual function may be used for presentation of environmental information, although not necessarily in a conventional way. Those with visual function at the level of n light perception" probably have significant temporal luminance modulation bandwidth that could be exploited for display of environmental informa- tion. If an individual has trichromatic light perception, the possibilities are significantly increased by providing color cues. RECOMMENDATION: Research should be conducted on novel ways of presenting information obtained from any visual information acquisition device to the severely impaired eye. Attempts should be made to permit the use of head movements in order to avoid the encumbrances of a hand-held device. Exploration should be made of the development of opto-electronic distance magnifiers and locomotion aids, given the existence of commercially available miniature video cameras and heads-up display technology being developed for pilots, to determine the usefulness of these new technologies to enhance the mobility of partially sighted people. These systems could help solve the problems of optical leverage with electronic damping of image translation and perhaps provide image processing tailored to an individual's impairment. Combination or Multimodal Displays Combinations of auditory and tactile and visual display technology may enable greater information display capabilities. A one-dimensional tactile display, for example, could be used to provide distance or direction information for nearby objects whose description is given auditorially (e.g. with a one-word verbal descriptor) after the manner suggested by Collins and Deering (1984~.
80 RECOMMENDATION: Combinations of display technologies should be explored, in particular combinations of visual displays with auditory or tactile input for partially sighted people and combinations of auditory and tactile input for blind people. Technology for Information Processing Contributions are anticipated from both signal-processing techniques and from selected subfields of artificial intelligence. Promising areas include: (1) Signal processing--e.g. image processing, including noise filtering, edge enhancement, and contrast enhancement and (2) Artificial intelligence--e.g. computer vision, including image analysis for object identification and contour following for map correlation; natural language, including voice input, synthesized speech output, and dialogue management; and inference mechanisms, including expert systems and script-based situation analysis. A final comment on the dynamics of technological development in our society is in order, for technology development has implications for research in the area of mobility for visually impaired people. Technology advance has sometimes been defined as a solution in search of a problem, and it is often cited as a primary cause whenever a technological device is produced that does not succeed in the marketplace. It is important to remember that technology advance is probably inevitable: it can be controlled but not eliminated. The only way to reduce its negative impact is through better theoretical and practical understandings of the problems in advance. And that requires that basic research on several facets of orientation and mobility be carried out. RECOMMENDATION: Research should be carried out to study the feasibility of incorporating contemporary and future electronic processing capabilities into mobility aids for the visually impaired to make devices n smarter" and easier to use.
