Interface for Adaptive Systems: Design of a Personal Wayfinder

Gitta Salomon, Yin Yin Wong, Tom Erickson, Joost Kemink
Advanced Technology Group
Apple Computer, Inc.

Apple Technical Report 117

Today's computerized devices are, for the most part, unaware of their users. They present the same images, the same interaction techniques and the same functionality, regardless of the characteristics of the person using them.

During the summer of 1992, our small, multidisciplinary research team undertook a design project confronting these issues. In the coming pages, we describe our process, some of the specific interaction techniques we developed, as well as some broader ideas about how industrial design (hard) interfaces and screen display (soft) interfaces might be integrated to achieve new product possibilities.
Wayfinding: A Domain for Exploring Adaptive Systems
Our approach to interaction design is user-centered: we choose a task domain,
study how people approach that task today, design a system suited to the
domain and its users, and then extrapolate from what we've learned to broader design principles. To better understand interface issues concerning adaptable systems, we chose a particular task domain - wayfinding - that allowed us to explore the concept of adaptation in several ways. It is important to remember that wayfinding is just one example of a task domain that might be supported by an adaptable system. By addressing the wayfinding task, we were able to explore and illustrate numerous ideas that could be useful in the design of any adaptable system.

What do we mean by wayfinding? Wayfinding has two components: figuring
out where to go and then getting there. Examples of wayfinding range from a child finding a new comic book store, to a Federal Express employee making a delivery, to a tourist touring Rome.

Today, a number of high technology companies are busily replicating traditional wayfinding tools such as paper maps and guide books in car-based navigation systems, or what we'll call wayfinders. However, it's not clear that they are taking the best approach. These systems present the user with precise, impersonal maps that include all the generic details found in traditional maps. Although these systems can keep track of their location in the world, they know nothing about their users particular needs or map-reading abilities. As we found through user studies, wayfinding is a personal task, one that is intimately entwined with the needs, preferences, abilities, and goals of the individual.

Wayfinding User Studies
As mentioned above, our approach to design is user-centered. This means that
it relies on an understanding of the needs and practices of those we envision using our technology. Two types of user studies formed the basis for addressing the design of a personal wayfinder.

In the first study, we asked people to draw maps as well as interpret maps drawn by others. The results from this study suggested characteristics that any wayfinder must possess. The second study focused on wayfinding needs that are specific to particular user groups. These findings suggested features that would support the wayfinding activities of each type of group.

In both cases, the studies were rather informal and the number of subjects small. Because our primary motivation was to develop new interaction ideas concerning adaptive systems - and not to completely design a wayfinding system - these types of studies were appropriate. In the past, we've found that such studies can quickly point the way towards new design ideas.
Study 1: Personal Navigation
The goal of the personal navigation study was simply to learn what is useful to people who are trying to find their way around. We chose a task that required people to draw and interpret (hand-drawn) maps because it seemed likely to provide insight into what was going on in their minds, and into how they used tools to support wayfinding.
We recruited 10 subjects from the Apple Employee Subject Pool. The subjects were both men and women, from a variety of backgrounds including finance, legal, administration, and engineering.

The study consisted of four parts:
Subjects were asked to provide directions to two familiar locations. In the first case, the subject was given a pen and paper and asked to show the experimenter how to get to his or her home from Apple. In the second case, the subject was told that the person he or she was producing the directions for was not present, but would be sent a fax. In this case, the task was to provide directions from a central Apple building to the subject's own office.
In the second part, the subject was given the "home map" drawn by the previous subject, and asked to verbally retrace the route for the experimenter (the first subject used a map drawn in a pilot version of the experiment).
Next the subject filled out a survey on wayfinding.
The study concluded with a general discussion of the drawing and interpretation tasks, the survey, and other relevant wayfinding experiences. Summary
The preferred method of giving directions was sketching, verbally annotating the
sketches, and jotting down textual shorthand labels as reminders. About 80% of the directions produced were in the form of maps with written annotations; the remainder of the directions were usually textual lists, though even these often had a small graphic component such as a sketch of an oddly-shaped freeway ramp.

We found that subjects were quite adept at interpreting rough maps drawn by people they had never met. Several features of the hand drawn maps appeared to facilitate this:
Roughness. Because it was obvious the maps were hand drawn, subjects were not mislead (for long) by distortions. The obvious roughness of the maps seemed to encourage them to apply their own knowledge in interpreting them.
Overview and detail. Subjects could simultaneously see overview and detail relevant to the particular route. Often the maps contained great detail at the beginning and the end of the route, with little detail offered in between (see Figure 1).
Landmarks. The presence of landmarks on virtually every map was clearly of great importance. Landmarks tended to be either street intersections or prominent features of the natural or built environment (e.g., steep hills; malls). A number of subjects pointed out the existence of personal landmarks (see Figure 2). A personal landmark is not often noticed by most people, but has significance to an individual or group. For example, one subject mentioned that he would use a certain shop as a landmark for his father-in-law, whereas that landmark would not make sense to anyone else.

Study 2: User Group Needs Assessment
This study examined needs specific to four types of user groups: adult professionals, children, the elderly and the vision impaired. Although each of these user groups requires support for the general needs outlined above, their particular abilities and deficiencies define how those wayfinding needs might be met, i.e., how a wayfinder might adapt to support them.

The needs assessment was a relatively informal process. One or two representatives of each user group were interviewed and asked about their general needs, the contexts in which they engaged in wayfinding, and ways in which wayfinding could be made easier for them. The following sections give a synopsis of each group.
Adult Professionals
The adult professional - with a wide range of abilities and interests - is the user that Apple typically designs for. In general, adult professionals are schedulebound and highly mobile. They have a wide variety of different, interleaved tasks, and they often need to be at a specific location by a certain time.
Because of the many tasks which are often entwined with wayfinding, adult professionals require a high degree of functionality. Their wayfinding needs extend over a wide region, and often include multiple forms of transportation.
Their needs vary from requiring directions to a specific destination to desiring to find a type of destination, e.g. a bookstore or a gas station. They often gather data about people and places they visit. For example, the business card exchange is a hallmark of the adult professional.

Donna, the professional adult we interviewed, described her needs when traveling to San Francisco. Since she doesn't own a car, she walks and takes trains and buses to get around. The use of trains to get her to and from the city require that she be able to return to the train station by a particular time. Her activities in the city may range from shopping, to visiting museums (relevant to her profession), to entertainment. Children
Children's wayfinding activities (when they are not in the company of adults)
are typically driven by leisure and socializing goals. In general, children's mobility is constrained to foot or bicycle travel, and they frequently find out about new places by word of mouth. Home and school are often used as reference points when navigating around their neighborhood, and often physical landmarks are more useful than street names.

We interviewed an eleven year old named Ben. Ben generally uses landmarks instead of street names to help him navigate. For example, he recognizes the intersection of Stevens Creek Boulevard and De Anza Boulevard as the crossing in front of the Apple building where his Dad works - not by the name of the streets. Ben's activities center on entertainment; comic book stores and playgrounds are areas of considerable interest. Ben typically finds out about new 'cool' places to go from his friends. For example, a friend recently told him about a new comic book store, and said it was near the karate dojo that they both attended.
The Elderly
In contrast to the adult professional, the elderly often have fewer goal-oriented wayfinding tasks. They are often not interested in finding a specific destination or meeting a particular schedule, but rather they wish to walk for leisure and exercise. However, health problems such as deteriorating vision, hearing and physical strength may constrain how far they are able to travel. Furthermore, they often have trouble using devices designed for a younger, more able-bodied population, due to these health problems. A tendency to become disoriented or forgetful can make them
reluctant to venture far. Feeling safe is an important issue. The ability to get help if overcome by illness or if they feel their physical safety threatened is of crucial importance.

We interviewed two elderly adults in a rest home. They often just want to walk for leisure and would like a route suggested based on the amount of time they wish to walk. Both seniors reported that they sometimes become disoriented: sometimes they
may not recognize normally familiar streets. Both seniors reported making modifications to their homes to help compensate for physical disabilities. One of the seniors made her microwave more usable by putting large numbers, written in black on white paper, in front of the buttons she could not see. She also used texture to help her identify the "off" state on her stove; by placing sandpaper at the correct position on the dial she can now feel when the knob is in the off position.
The Visually Impaired
Blind or visually impaired adults obviously have special needs for wayfinding support. Ideally, a visually impaired adult should be able to travel to and from work or school, stores, and entertainment. There are a number of physical aids
that help the visually impaired in navigating today. Braille is an effective way of presenting textual information, though reading speed tends to be about 28
words per minute, rather than the couple of hundred words per minute
supported by visual reading. Some tactile displays that present Braille characters in a manner similar to electronic scrolling signs exist. Although the visually impaired can use paper-based tactile maps to help them find their way, finding the correct orientation of the map to the world is a particular challenge.

We interviewed two visually impaired adults. They pointed out that while they
obviously can't use visual landmarks, tactile landmarks (path texture, air flows), small objects (e.g., fire hydrants), sounds (e.g., major streets, viaducts), and
smells can assist with orientation. Designs for a Personal Wayfinder
Inspired by the informal user studies, we began examining designs that would support the construction of wayfinders suited to the needs of particular user groups. We arrived at a two part strategy that 1) separates the generic aspects from the personalized aspects of a wayfinding system and 2) uses both the
hard and soft interfaces to promote adaptation.

A Strategy for Creating Adaptable Systems
The Base: Encapsulating Generic Functionality
Today, our computational systems are relatively unaware. However, a system could become more aware of its surroundings through various sensors. In particular a device could be cognizant of its whereabouts, its bearing and its orientation. Location-sensing technologies, such as the Global Positioning System, exist today and makes such knowledge available. Imagine, then, that we package the sensors necessary to deduce this information together with a special purpose CPU optimized for map data and route calculation. We then possess the computational technology - in one piece of hardware - necessary to support the generic wayfinding task. We will refer to this item as the wayfinding 'base.'
The Clothing: Supporting User Group Adaptation
As we found in our user studies, different user groups have different wayfinding needs. These needs could be described as ways of biasing the base
technology, of adapting or tailoring it to support particular tasks. Essentially, we need to represent the core functionality of the base differently to different user groups. In order to do so, we developed a design that literally and figuratively masks the base. In this design, the base is covered with a hard interface - or 'clothing' - that becomes the part of the device the user interacts with directly (see Figure 3).

By decoupling the core technology from the clothing, different user groups can
be accommodated. For example, while the base technology supports the
generation of maps, the clothing can support the best style of display (e.g., a tactile, rather than a visual display, for the visually impaired user). One type of clothing can hide inappropriate functionality from the child or novice, while another type of clothing may make it available to the more sophisticated user.

By snapping a particular clothing style onto the wayfinding base, users tell the system something about their predilections; the system immediately becomes
more attuned to their wayfinding needs. However, while the clothing can get the device into the right range of functionality quickly, it does not make the wayfinder a completely personal device.
The Soft Interface: Adapting to the Individual
A fully adaptable wayfinder would provide an interface and functionality appropriate for a specific person. For example, over time, the wayfinding device could observe and note the personal idiosyncrasies of its user (such as which landmarks are particularly meaningful) and proactively adapt to become more useful to that individual. Alternatively, a user could explicitly tell the system about personal preferences. We foresee that either style of individual personalization will occur through the software-controlled screen interface. For example, the clothing may select the type of representation (a visual display), but the individual's preference for spatial maps with 'home' as a landmark, instead of text-based directions emphasizing the distance from work, would occur through the soft interface.
Applying the Strategy
Assuming a standardized, generic wayfinding base, we designed four distinct wayfinding systems that support each of our four user groups. In each case, industrial designs for clothing and on-screen soft interfaces were conceptualized. These designs were executed to varying degrees, but were complete for our purposes; our intent was to adequately explore the base and clothing concept, and not to fully develop working prototypes.

A few themes are common to each of the designs. Each system is capable of keeping the map display oriented to the physical world, showing the user's current location, and creating personal landmarks for the user (although the nature and method of creation differs from system to system). In the designs
incorporating visual displays, we use a round display. This design decision allows the user to maintain the same amount of context around a map location
regardless of the device's orientation. In each case, the system constructs routes based on the user's specification of a destination or goal.

The following sections provide insight into the designs developed.
Design Overview: The Adult Professional's Wayfinder
Stylistically, both the hard and soft interfaces for the adult professional's device have a clean and elegant appearance (see Figures 4 and 5). This wayfinder utilizes a touch screen and stylus to provide greater flexibility for browsing and creation of data. Since adult professionals are highly mobile, the wayfinder allows them to navigate within small or large regions by changing the scale of
the map or by panning the map in any direction.

The wayfinder allows the user to view a map with different overlay 'filters' that emphasize items of particular interest (e.g., only clothing stores, or only galleries, within a certain vicinity). The size of the filtered region changes based on the amount of time the user wants to spend traveling and the mode of transportation used.

The wayfinder can help goal-directed professionals by giving them the means
for constructing a route to a specific destination or general area. An overview and a detail view of the route are displayed simultaneously; a user can attend
to local navigation needs while maintaining context of the entire route. The map aligns to the world and adjusts to show current position and orientation as the device moves.

Professionals who are schedule-bound can leave a 'timer landmark' - a landmark to which they must return at a specific time. The wayfinder keeps track of the
user's position, relative to the timer landmark, warns the user when it is time to start the return trip, and posts a route of return from the current location.

Finally, professionals commonly take business cards from destinations they wish to remember. These users often then spend substantial time organizing this
information for later retrieval. The adult professional's wayfinder allows users to collect data from locations using wireless communication. (This, of course, assumes a certain amount of infrastructure; a business would have to broadcast this information). In this way, individualized landmarks are established and important information is collected for later use or sharing.
Design Overview: The Child's Wayfinder
The clothing and soft interface for the child's system uses simple graphics, primary colors, and auditory feedback to reinforce a familiar, game-like interaction (see Figures 6 and 7).

This wayfinder's functionality and interaction style are simplified.The child interacts with the system via a touch screen. The wayfinder includes access to
the phone directory; it is filtered to only include items of interest to a child. Since the child's mobility is limited, the wayfinder allows the child to browse within a local vicinity. Again, functionality is pared down, maps can be viewed only at 3 different scales while the professional's device supports multiple scales. Because user interviews indicated that home and school often act as reference points for children, the wayfinder provides hard buttons to quickly view destinations or the current location in relation to these familiar landmarks.

To visit a new place, a child can request a route to a specific destination. An
overview of the route is provided in the form of a simplified, variable scale map. This map is similar to the maps users drew for one another in the map study. For example, greater detail is provided at the beginning and end of the route; personal landmarks along the path give context based on what the child

Since children approach tasks casually, they are presumably not motivated to spend time setting up their own personal systems. Therefore, inference on the machine's part may play a greater role in the child's system. For example, instead of requiring the child to enter landmarks, the system can infer personal landmarks based on the duration spent at any one location. A landmark will be
reinforced with repeated visits, but if a place is never revisited the landmark will gradually fade away.
Design Overview: The Elderly Person's Wayfinder
The wayfinding system designed for the elderly incorporates high-contrast
colors, textured buttons and a simple, easy to read display to support legibility (see Figure 8). In this system, the underlying functionality of the base is used to help an elderly person construct a walk around the neighborhood of specified duration. The user can indicate the desired duration using a familiar timer-style indicator knob and the system will suggest a suitable path. The user can choose
to see a simplified overview of the path or a close-up view of the current location and the immediate surroundings.

The user interviews indicated that the elderly, in particular, are security conscious. Several features were included to help these users feel more secure about going for a walk. For example, while setting a duration the system will inform users if the requested walk will keep them out after dark. At any time the user can choose to be routed home from their current location. In addition, a button that alerts others to an emergency condition is provided. Elderly users are reminded to wear this button before using the device; it must be detached before other controls can be operated.

The clothing is also designed to support areas where these users are physically challenged. For example, it provides a grip that is suited to reduced arm strength. By gripping the handle with the palm and resting the device along the arm, the weight is distributed, therefore requiring less strength to support the
device (see Figure 9). There are only a few buttons, and their recognizability is enhanced in several ways. For one, the buttons use high-contrast visuals such as complementary colors to provide visual clarity. They are also tactile; by applying a unique texture to each button surface, additional feedback for those with poor vision is provided. Finally, as yet another way of compensating for poor eyesight, the buttons describe their functionality via recorded speech as a finger runs over them.
Design Overview: The Visually Impaired Person's Wayfinder
The clothing for this wayfinder consists of two parts: a small, hand-held device that the base slips into (Figure 10) and a headset with speaker and microphone
(Figure 11). In place of a visual display, the clothing provides a tactile display (an array of pins) that is finger-sized. The hand-held device is designed to fit in the user's palm, with each of the fingers conveniently placed over each of the buttons and the tactile display. A user can depress the button directly under the middle finger and speak a desired destination. The system directs the user to this destination via the tactile display.

In place of visual landmarks, the visually impaired person uses auditory, tactile, and olfactory landmarks to navigate an area. This wayfinder allows the user to create personal landmarks by speaking a landmark description while depressing
the 'landmark-making' button located under the thumb. The audio recorded is played back the next time the user arrives at the location, as sensed by the base. Because visually impaired users cannot see their current context, one of the most powerful functions of a personal wayfinder is to allow them to understand what is in the vicinty. By pointing the device in a certain direction and depressing another button the user can hear, in order, the items located in that direction, including personal landmarks that have been previously recorded. The user might hear, for example, that a left turn would take him past a large intersection in three blocks, followed by a drug store and then a coffee shop on the fifth block.

In Conclusion
The users we interviewed about map making and reading strategies weren't concerned with precision and perfect scale in maps. Rather they cared more about appropriate emphasis and personal landmarks. While most computerized mapping systems concentrate on providing exact, deliberately unbiased maps,
we explored creating 'user-readable' maps. These maps are not exact representations; they are are loosely drawn and include personal, idiosyncratic landmarks.

Our interviews with representatives of different user groups allowed us to
informally gather information about possible differences in their wayfinding needs. We realized that all of the users could benefit from the same basic functionality, but would want to access and use it in different ways. We therefore devised a design strategy that divides the elements of the system into a 'base,' 'clothing,' and a software interface.

This strategy was used to create four designs which led to general insights about the design of adaptive systems. We found that completely different representations of the underlying technology might make a device suitable for different types of users. The elderly person's wayfinder presents very limited functionality compared to the other systems, and yet it addresses an important need for these users. Because the base includes a clock and some knowledge about temporal events, the device can inform the elderly that a walk will keep them out past dark. This same technology allows professionals to set a timer landmark.

The role the user plays in the system's adaptation can vary with the task at hand and the type of user. As noted, children's landmarks are inferred by the device because a suitable means exists to do so, and because children are unlikely to do so themselves. On the contrary, adults can intentionally create landmarks by collecting data of interest at specific locations.

The ways in which a device communicates should also be determined by the
task and type of user. Obviously, a tactile and audio based system makes sense for visually impaired users. But for users whose hands are busy - e.g., mechanics - audio input and output is also the best means of communication. Although technological advances might make it possible to provide all input and output modalities in every situation, this may not always be desirable.

Incorporating core technology in a base, and then adapting that technology for
different users through clothing, is a flexible strategy that has applications in numerous domains. For example, wholesalers, shelf stockers and shoppers
could all use a device whose base contains inventory and related data, and whose clothing presents this information in different ways. A student could use a computational device at school with one set of clothing geared towards
learning (a new interpretation of the school uniform!) and use a different set of clothing at home for playing games. Clothing could also adapt a device to situations requiring different degrees of ruggedness; a lightweight covering might suffice for office use, while a robust exterior could be used in an industrial environment.

As we've often found, a user-centered process, focussed on a specific task domain can lead to innovative designs that both address the specific domain as well as broader contexts.
This project took place under the auspices of the Human Interface Group within Advanced Technology at Apple Computer. Special thanks are extended to several
others who made the work possible. Norio Fujikawa and Eduardo Sciammarella joined the team as industrial design interns and were responsible for creating the various clothing and base models shown in this paper. Ian Small also contributed to the development of the clothing and base concept. Sally Grisedale assisted with the preparation of this paper.

See Also