Individuals with disabilities have problems accessing computers because of keyboards, mice and monitors. But these devices only come into play when computers communicate with people. When computers communicate with each other, keyboards, mice and monitors aren't involved. This suggests a way of separating the problem of access into two. First we provide an individual who has a disability with complete access to one computer. Then we provide them with access to any computer, by letting the one computer to which they have access take over the job of communicating with the rest. The Total Access System is based on a this separation.
The Total Access System was initially conceived
by Neil Scott
, and key pieces of the
system have been designed and implemented by him and others
at the Archimedes Project at Stanford University (see
box).
This separation is embodied in the Total Access System in
its two main components, the Personal Accessor and the
Total Access Port or TAP. Personal Accessors vary
from person to person according to the user's abilities and
preferences. TAPs link the Personal Accessor to any
host computers that the user wants to work on. The Personal Accessor and the TAP
communicate with each other in a high-level functional
language we call the Archimedes Protocol.
A Personal Accessor is the conceptual solution to the first set of issues: providing an individual with access to one computer. A Personal Accessor is a personal computer with the hardware and software for the accessibility devices that a particular person needs built into it. A quadriplegic, for example, would have speech and head-pointing or eye-tracking built into his or her accessor. A person with advanced ALS would have eye-tracking, but not speech or head-pointing (since he cannot use these). A blind individual's accessor might include a speech synthesizer or a tactile display.
An accessor can be made small and portable and can travel with the user in the same way that many people now carry palmtops and laptops. It can be exactly tailored to the individuals needs and preferences, containing what is needed and not using memory or physical space for things that are not needed. Its only function is to provide access. Thus it does not become obsolete when host operating systems or applications are changed. Because it is modular, it can be easily upgraded as access tools improve. It separates input and output functions from applications so that it provides a consistent interface across devices and applications. It allows access to any host and computer-driven technology outfitted with TAPs including kiosks and microwave ovens.
A TAP is the conceptual solution to the second set of issues: access to any computer. A TAP attaches to a host computer through the keyboard and mouse ports. The connection between Accessors and TAPs is through a (wire or wireless) link that uses a specially developed serial communications protocol that is independent of both the accessor and the host. Standardization of the protocol allows any accessor to operate with any host device. From the perspective of the host, an accessor is indistinguishable from standard I/O devices. Input from the accessor through the TAP emulates a keyboard and mouse; output from the host computer through the TAP is displayed by the accessor in a manner appropriate to the user.
Because the TAP fools the host into thinking that it is getting its own keyboard and mouse input, Personal Accessors work on all applications and interfaces. Some alternative input strategy is used to ``press'' the key and ``move and click'' the mouse. It might be voice, Morse code, or single switch scanning. No matter what is used, it is all the same to the host computer; it is interpreted as keyboard and mouse input.
In this sense, accessors work equally well for all applications. Accessors can be made to work more efficiently for a given application by means of user-defined macros that are specific to the task and the way the user likes to work.
The TAP keeps the adaptive work outside of the host and therefore doesn't interfere with the functionality or speed of any of the applications running on the host. It is small (currently slightly larger than a computer mouse), relatively low cost, and simple to install, all of which encourage widespread access adaptation.
Future TAPs will also collect control signals, raw
text, raw video, and raw sound from the host and transfer
this information
back to the Accessor for processing into a form that is
accessible to the user. This is a crucial piece of the
solution to the Graphical User Interface problem for blind
computer users.
Consider Jorge, a quadriplegic who uses his voice to control his computer. While at work Jorge's accessor is usually connected to a Macintosh desk top computer. Jorge speaks into a microphone. His words are recognized by a voice recognition program running on his accessor. The intensive memory and CPU demands of the voice recognition program do not affect Jorge's Macintosh. Jorge's accessor contains a software shell that allows Jorge to use intuitive macros suited to different applications, for example, he might say ``begin fax'', ``read mail'', ``replace word'', ``spell checker'', ``print file'', etc. The data from the accessor bypasses the keyboard and mouse of Jorge's Macintosh. The Macintosh-TAP converts the data to virtual keystrokes and mouse movements--Jorge's Macintosh cannot tell that it is being controlled by voice rather than by fingers.
Later in the day, Jorge needs to use a Sun workstation. The accessor stays the same; the macros are the same. The Sun TAP converts the data from the accessor to signals that supply virtual keystrokes and mouse movements to a Sun.
Jorge takes his accessor with him when he goes home. There
he could in principle use it not only to operate his home
computer, but also to operate his television, stereo, VCR, microwave, and so forth. The TAS design concept
would apply equally well to kiosks and ATMs. Kiosks
and ATMs outfitted with TAPs would be accessible
to everyone with an accessor.
The TAS strategy is contrasted with the traditional ``in-host'' strategy: locating the assistive hardware and software in the host computer that the individual with a disability uses. The basic advantage that the TAS offers over the in-host strategy can be seen as roughly the difference between addition and multiplication. On the TAS strategy, one has to develop an accessor that outputs the Archimedes protocol for each input device and develop a TAP that inputs the Archimedes protocol, for each type of computer. To compute the number of technological problems, we add the number of assistive devices to the number of types of machines. With the in-host strategy, each combination of input device and type of host constitutes a separate problem. To compute the number of technological problems, one needs to multiply the number of input devices times the number of hosts.
There are further advantages, too. First, and foremost, TAS isolates the user from the whims of hardware and software designers. The accessor interfaces to host computers through a TAP which emulates the electrical operation of the physical keyboard and mouse. Keyboard and mouse functions are fundamental to computing and manufacturers get little advantage from changing them in anything but purely cosmetic ways. The IBM PS/2 keyboard and mouse, for example, is becoming a de facto standard throughout much of the computer industry, even among competing products. The ubiquity of the keyboard and mouse makes them a point of stability in an otherwise constantly changing world. They have become part of the infrastructure. Variations to the mouse, such as trackballs, finger pointers, touchpads, and the like, all use the same electrical protocols as a basic mouse. TAS currently supports TAP interfaces to IBM PC, SGI, Macintosh, and Sun computers. Any computer based device or appliance can be made accessible by connecting a suitable TAP. The TAP becomes part of the infrastructure.
Another significant advantage of TAS is that it allows a disabled person to use a single accessor to operate any computer or device that has been equipped with a TAP. A properly chosen and configured accessor provides a disabled user with a very high level of independence and will last a very long time. It therefore makes good economic sense to invest whatever it takes to match an accessor to the needs of a disabled individual.
Research at the Archimedes Project has shown yet another real advantage of using the TAS design. Many different technologies are potentially useful for disabled individuals. Developing the necessary hardware and software interfaces to real-world tasks, however, is usually a far from trivial exercise and many good ideas languish due to the effort required to evaluate them in a real application. The TAS provides an ideal vehicle for evaluating and incorporating new technologies because it automatically connects new access devices directly to the existing infrastructure. An eye-tracker, for example, follows the movement of a user's eye and generates a stream of data showing where the user is looking. With suitable software, the eye-tracker can be used to emulate a keyboard or mouse. The question is, which keyboard and mouse should it emulate? This question is moot if we configure the eye-tracker as an accessor. The researcher need not be concerned with what the eye-tracker is to be connected to since anything that can be controlled by a keyboard or mouse can be controlled by the eye-tracker.
The TAS allows several different accessors be used simultaneously on the same host system. This leads to several interesting possibilities. For example, more than one person can have equal access to a single host device and can therefore work cooperatively on a single project. Similarly, a single user can operate several different accessors at the same time and can therefore mix and match different input strategies to suit the tasks being performed. One very effective example of this is the combination of speech recognition and head tracking. The speech accessor handles all text input, program commands, and pressing or clicking of the mouse buttons, the head tracker handles all of the pointing functions. The combination of speech recognition and head tracking is significantly more effective than either technology used by itself.
It is a small step to see that the TAS concept has advantages for non-disabled individuals whose experiences, preferences, or work-conditions may dictate or encourage one type of access over another. For many professors, executives, physicians, and lawyers, for example, talking is easier and faster than typing. Individuals who drive and use cellular phones will need to talk rather than type to their computers. Car radios would be safer if we could operate them with our voices instead of with our eyes and hands. Employees of the tele-marketing industry would be more productive if the spoken words used to confirm addresses and orders could simultaneously enter the data into the computer database. ATMs that could be operated by voice from within one's car would be popular with everyone. As John Thomas puts it in an article in this volume, accessible issues ``force designers to think out of the box.'' He goes on to make the important point that when the communication system is made accessible to individuals with disabilities, everyone gains access to those individuals. He says, ``Providing access for people with special needs is not just for them--it's for everyone.''