Designing a Home of the Future - MIT Media Lab

Page created by Harvey Goodwin
 
CONTINUE READING
I N T E G R AT E D E N V I R O N M E N T S

                 Designing a Home
                 of the Future
                 An interdisciplinary team is developing technologies and design strategies
                 that use context-aware sensing to empower people by presenting information
                 at precisely the right time and place. The team is designing a living laboratory
                 to study technology that motivates behavior change in context.

                                P
                                        eople spend more time in their homes than in      no single “home of the future.” In particular, we aim to
                                        any other space. The home ideally provides a      create design strategies for more flexible environments that
                                        safe, comfortable environment in which to         better meet occupants’ physical and cognitive needs than
                                        relax, communicate, learn, and be enter-          current environments. Based on discussions with medical
                                        tained. Increasingly, it is where people con-     professionals, patients, educators, and homeowners, we
                           nect with friends and family, conduct business, manage         believe that the home of most value in the future will not
                           resources, learn about the world, and maintain health and      use technology primarily to automatically control the envi-
                           autonomy as they age. People invest extraordinary              ronment but instead will help its occupants learn how to
                           amounts of time, money, and emotional energy to mold           control the environment on their own.
                           their homes into living spaces that meet their needs.             This shift is the focus of this article. As a byproduct of
                              Unfortunately, homes today are ill-suited to exploiting     this shift, new tools are required to study technology in the
                           the pervasive computing applications being developed in        context of home life. To address this need, our team has
                           laboratories. Most homes do not easily accommodate             designed and is planning to construct a “living laboratory”
                                                 even the simplest new technologies,      that will support qualitative and quantitative studies inves-
     Stephen S. Intille                          let alone embedded sensor infra-         tigating the relationships between spaces, the behaviors of
     MIT School of Architecture                  structures and ubiquitous display        people, and pervasive computing technologies.
     and Planning                                technologies. Moreover, homeown-
                                                 ers generally believe that computer      Envisioning homes of the future
                                                 devices make life more complex and          If we are to believe most movies, television, and pop-
                           frustrating rather than easier and more relaxing. They are     ular press articles that mention home life in the future, we
                           wary of the aesthetic, financial, and cognitive challenges     will have complete control over our spaces at the touch of
                           of bringing new technologies into their homes.                 a button. In fact, our homes will be so fully automated
                              Researchers in the Changing Places/House_n: MIT             and “smart” that we will rarely have to think about every-
                           Home of the Future Consortium (http://architecture.            day tasks at all. We will spend nearly all our time in the
                           mit.edu/house_n) at the Massachusetts Institute of Tech-       home engaged in leisure activities because digital and
                           nology are investigating how the home and its related tech-    robotic agents will have taken over the mundane chores
                           nologies, products, and services should evolve to better       of day-to-day life.
                           meet the opportunities and challenges of the future. Our          Researchers and technologists are more cautious in pre-
                           team’s researchers have backgrounds in computer science,       dicting the future of the home. A survey of ongoing work
                           user interface design and usability, architecture, mechani-    shows, however, that there is a bias toward creating auto-
                           cal engineering, psychology, and materials science. The        matic (smart) home environments that eliminate the need
                           “n” in House_n represents a variable; we believe there is      to think about tasks such as controlling heating and light-

80   PERVASIVE computing                                                                                          1536-1268/02/$17.00 © 2002 IEEE
ing, going to the grocery store, scheduling home    mize savings and comfort by automatically con-          when making these control decisions. Because
appliances, and cooking.1–3                         trolling the HVAC systems, windows, and                 the system is so complex, the user will be left
    Although our team might use automation to       blinds. For instance, on a day when the tem-            feeling frustrated—helpless to understand the
help people accomplish tasks they cannot per-       perature is predicted to shift from warm to cool,       behavior. Why does it keep opening the win-
form on their own because of a disability or        the home might determine that the optimal               dows when, clearly, the user wants and needs
frailty, our primary vision is not one where com-   cooling strategy is to shut down the AC and             them closed?
puter technology ubiquitously and proactively       automatically open a set of blinds and windows
manages the details of the home. Technology         so as to create an efficient cross breeze.              The home that uses subtle reminders
should require human effort in ways that keep          This scenario is relatively simple compared            Consider an alternative scenario. In this
life as mentally and physically challenging as      with other smart-home visions. In practice,             home of the future, the windows include a tiny
possible as people age. We are designing and
building prototypes demonstrating how to cre-           Technology should require human effort in ways
ate environments that help people
                                                                 that keep life as mentally and physically
• Live long and healthy lives in their homes
• Reduce resource consumption                                      challenging as possible as people age.
• Integrate learning into their everyday activ-
  ity in the home                                   however, it would be an immense challenge to            light that is either embedded in the window
                                                    achieve this simple scenario in an actual home          frame (for example, a light-emitting diode) or
   To accomplish these goals, we envision com-      setting. The sophistication of commonsense              projected on the window using display tech-
puter technology as ever-present but in a more      reasoning and context awareness that is                 nology (for example, an IBM Everywhere Dis-
subtle way. Information will be presented to        required is daunting, given the current state of        play6). The home’s embedded sensors and opti-
people at precisely the time and place they need    our understanding of these fields. There are            mization algorithms compute a strategy for
it. We want our pervasive technologies to           many situations in which the automatic system           cooling the home by opening a particular set of
empower people with information that helps          might succeed in optimizing temperature com-            windows, but they do not proactively imple-
them make decisions; we do not want to strip        fort yet fail in “doing the right thing”: some-         ment the strategy.
people of their sense of control, which has been    thing noisy is occurring outside, someone is               In this example, imagine that the light on the
shown to be psychologically and physically          smoking outside the window, someone in the              window subtly illuminates. It does not inter-
debilitating.4                                      home is allergic to pollen and the pollen count         rupt the home occupant. When someone in the
                                                    is high, it is raining outside, it is too quiet for a   home notices it, he or she knows the light means
Control versus empowerment:                         person reading when the hum of the air condi-           “it might be a good idea to open this window
An example                                          tioner is off, someone did not want the blinds          right now.” The home thereby unobtrusively
   To illustrate this shift in thinking, imagine    open because it throws glare on a computer              informs the user of actions that might be taken
that our goal is to create an environment that      screen, and so on. No matter how hard the sys-          to conserve energy or money. In this way, the
uses pervasive computing technology to save         tem designer tries to program contingency plans         home teaches the occupant, in an unobtrusive
energy by automatically controlling the heater-     for all possible contexts, invariably the system        way, how to achieve the optimal settings. The
vent-air conditioning system. We assume that        will sometimes frustrate the home occupant              home can take a similar approach when the
the environment’s embedded sensors can infer        and perform in unexpected and undesirable               goal is to improve health or introduce learning
context such as where people are, what they are     ways. A learning algorithm would also have              into everyday life.
doing, and what the inside environmental con-       difficulty because a training set will not con-            This scenario has several advantages over
ditions are. We also assume that the home con-      tain examples of appropriate decisions for all          proactive control:
tains computer-controlled HVAC appliances,          possible contextual situations.
windows, and blinds.                                   There is a fundamental problem here: the             • Information can be presented that the occu-
                                                    more complexity the algorithms consider when              pant can react to without interrupting ongo-
The automated home                                  making decisions, the less transparent those              ing activity in potentially irritating ways; this
   One way to reduce resource consumption is        decisions will be to the homeowner.5 The system           is especially true if information can be “aug-
to design a home environment that controls          will actually become less predictable as it               mented” onto the physical environment itself.
environmental conditions. The home’s occu-          acquires more expertise, and the system’s suc-          • Leaving occupants in control of making deci-
pant informs the system via some type of user       cess some or most of the time will raise user             sions allows the home to present options
interface that he or she wishes to stay comfort-    expectations about what the system is capable             based on partial information without con-
able while saving as much energy or money as        of doing. Inevitably, the system will violate the         fusing them; they will naturally consider con-
possible. The home then uses a set of opti-         user’s high expectations given the unexplain-             texts that the home has not and adjust their
mization algorithms to simultaneously maxi-         able “intelligence” the system sometimes shows            actions accordingly.

APRIL–JUNE 2002                                                                                                               PERVASIVE computing                 81
I N T E G R AT E D E N V I R O N M E N T S

     • Algorithms that make suggestions can                 decides to take an action such as opening the         occasionally follow the home’s recommenda-
       degrade gracefully; algorithms that make             window. This is a “point of behavior” that the        tion, they will gradually learn how to efficiently
       decisions typically do not.                          home can easily identify by detecting a specific      control the temperature in the environment in
     • Lack of control over aspects of life has been        event (the opening of the window). The user           sophisticated ways. Occupants will understand
       shown to diminish health4; this strategy             has already decided to stop whatever he or she        that lights appear on their windows because it is
       empowers the home occupant.                          was doing to perform a recommended task.              cool enough outside to set up a cross breeze.
                                                            The home can safely infer directly from sen-          They will also gradually learn how to create a
        The user ultimately decides whether to open         sor data that the user is opening the window          cross breeze given the geometry of their house
     the window. Therefore, the task of interpreting        and therefore is likely to be receptive to infor-     and the prevalent wind direction:
     the suggestion in context rests with the user: if it   mation that helps the user determine how to
     is noisy outside, the user will simply decide not      do so. The user is also likely to be curious          • Using window inlets and outlets that maxi-
     to open the window, realizing that this is not a       about why the home is making this recom-                mize cooling and air flow through the home
     good time. This is a pervasive computing appli-        mendation. Finally, the user will have moved          • Understanding how long it will take for cool-
     cation with an exceptionally simple user interface.    to the object’s physical location; this presents        ing to occur
     Could such a system actually influence behav-          a good opportunity to teach by overlaying dig-        • Recognizing the best times to establish intake
     ior? Yes. Controlled studies in homes show that        ital information on the physical space. In this         air
     using such a small, simple light on an AC unit can     scenario, precisely when the system determines        • Knowing how to use fans to facilitate cross-
     lead to 15 percent reductions in AC use.7,8            that the user has decided to act, it can overlay        breeze cooling
                                                            information on the real world to educate the
     The teaching home                                      user about how to create the most effective           Most people do not know how to do these things
        Pervasive computing can be used not only to         cross breeze.                                         because no one is there to teach them when they
     motivate behavior but to teach at the moment              Even if the user does not have time to stop        need guidance at the point of behavior. A perva-
     when the behavior is undertaken. Systems that          and study information, it is possible to present      sive computing system presenting information
     automatically make control decisions generally         feedback that results in learning. For example,       at the point of behavior can fill this need.
     miss this opportunity—users can become com-            as the window is opened, the system might pro-           To measure the impact of point-of-decision
     placent if the system functions perfectly.             ject information onto the nearby wall that esti-      messaging, we are developing prototypes on two
     Although a computer system might try to pre-           mates the magnitude of the breeze to be created.      platforms: “augmented reality” Everywhere Dis-
     sent the user with educational messages to             The person might notice that, counterintuitively,     play technology that can place information
     explain the actions it is taking, to do this with-     opening the window further does not always            directly onto objects in a home6 and portable
     out interrupting and irritating the user is a chal-    result in a stronger cross breeze. The user’s task    computing devices such as PDAs and cell phone
     lenge. The system must compute a reasonable            has not been interrupted, so even if the user is      hybrids. (Based on trends in miniaturization,10
     time to present the information. Even for rela-        completely uninterested in the information, no        within five to 10 years we will likely be able to
     tively simple help applications, this has proven       attentional disruption has been created. Imme-        buy an affordable and stylish wrist computer
                                                                                                                  that will convey information to us whenever it
             Pervasive computing can be used not only to                                                          is appropriate.) Either method can easily pre-
                                                                                                                  sent information at the right place. The chal-
           motivate behavior but to teach at the moment                                                           lenge then becomes to develop algorithms that
                                                                                                                  can recognize the right time and select a pre-
                        when the behavior is undertaken.                                                          sentation strategy suitable for the given context.
                                                                                                                     We have been conducting small user studies
     to be difficult to do (for example, ClipIt, the        diately after the point of action, the system         with mock-up displays and are now imple-
     Microsoft paperclip, attempts just-in-time help        could remove the information.                         menting prototypes of some of the examples.
     but does so in ways that often require the user                                                              Figure 1 shows two displays that might appear
     to divert attention from the current task). On         Our project                                           on a wearable PDA device and convey the same
     the other hand, if a user is unhappy with a con-          It is the potential impact of this nonintrusive,   type of information in different ways. An open
     trol decision the home has made, he or she will        just-in-time learning on behavior that our group      question that we plan to explore is how the pre-
     feel annoyed and primarily interested in coun-         has begun to explore. We are interested in three      sentation of information and the current con-
     teracting the home’s actions. This is not the best     points in time: the point of decision, the point      text influence the persuasive impact of educa-
     time for the home to present explanatory infor-        of behavior, and the point of consequence.9 How       tional messages.
     mation to promote learning. A home that leaves         can we use sensors that automatically detect             An important consequence of using such
     control to the occupant avoids this tricky issue.      these specific (and sometimes fleeting) moments       technology for just-in-time teaching rather than
        The extraordinary potential power of per-           in time to educate people about controlling their     control is that the information people learn is
     vasive computing comes into play when a user           environment? In this example, as the occupants        transferable to other environments where there

82   PERVASIVE computing                                                                                                          http://computer.org/pervasive
Open the marked windows now:                                                                                  Opened this week
                                                                               Turn AC on for 1 hour
         *                                                                        (windows closed)

                                                                               Then open these windows:
                                                                                  Bedroom(full)
                                                                                  Kitchen (halfway)
         *
     And open the front door.
     Wind:     NW @ 10 mph                                      WHY? It’s too warm outside                             You        10 closest neighbors
     Temp:     74°F inside, 78°F outside                                Pollen count high                          2.3 minutes        2.1 minutes
     Forecast: Cooler breezes from NW this afternoon            But in an hour temp. will drop
                                                                                            Another
             CANCEL                OK                                                        option

   (a)                                                    (b)                                                (c)

Figure 1. The messages in (a) and (b) might appear on a person’s PDA device the moment he or she opens or closes a window. The
message in (c) could be displayed on the refrigerator or on a wearable wrist computer just as someone closes the refrigerator door.

is no computer technology. Additionally, the           serve as a “living laboratory” to study how peo-         We are developing a component-based build-
just-in-time teaching scenario might still use         ple live with technology. Applications will use       ing system for new homes that will be used to
automatic control of the windows but in a way          this architecture to acquire information about        construct the living laboratory. The flexible
that encourages people to use their physical abil-     context. One existing living lab, the Georgia         nature of the methodology will let us transform
ities: a young healthy person would be encour-         Institute of Technology’s Aware Home, is being        the lab’s physical and digital infrastructure
aged (using ubiquitous messaging) to exercise          used for such studies.1 We will use our lab infra-    quickly at low cost to study different spatial and
muscles by opening the window, whereas a frail,        structure to quantitatively measure and quali-        sensor configurations. We will use the lab to
elderly person who cannot lift the window              tatively study the impact of new technologies         identify the technologies that are most suitable
would be encouraged to go to the window and            on people’s behavior in a real living environ-        for retrofitting existing environments.
push a real or virtual automatic button. We are        ment. This facility will not be a traditional home       Eighty percent of the cost of building a new
also studying how to present persuasive, perva-        that has been retrofitted with technology but         home in the US is spent on field labor and 20
sive messages to motivate small behavior               an entirely new type of home structure designed       percent on materials. We are interested in revers-
changes during everyday activities. For instance,      from the ground up to serve three functions:          ing this ratio so that four times as much money
Figure 1c shows a message that could be dis-                                                                 can be devoted to materials, design, engineer-
played on a refrigerator door (or on a wearable        • Demonstrate a new type of building method-          ing, safety, and technologies in the home. Bor-
wrist computer) just after the door has been             ology that lets us embed technology within          rowing from recent innovations in the automo-
closed to encourage awareness of energy con-             the infrastructure of environments and then         bile industry, House_n researchers led by project
servation. People who are informed that their            easily change and upgrade it.                       director and architect Kent Larson have pro-
behavior is out of line with community stan-           • Provide an environment in which to scien-           posed an integrated “chassis–infill” construc-
dards will often naturally change their activity;        tifically study home life, particularly the rela-   tion system that can be rapidly installed with
in this case, a greater awareness of the need to         tionships between space and information.            minimal labor. In one integrated assembly, com-
keep the door shut might result.                       • Provide a means for evaluating whether new          posite beams and columns provide structure,
                                                         types of pervasive computing interventions          insulation, sensor arrays, lighting, signal and
A living laboratory                                      have a long-term and meaningful impact on           power cable raceways, and ductwork. The
   Our team’s focus affects not only the type of         behavior in the home.                               beams use special connectors that lock together
technology we are designing but also our out-                                                                easily. Infill sections that form the structure’s
look on how we must conduct research to eval-          Demonstrating a building methodology                  interior and exterior walls are then “snapped
uate our work. We cannot evaluate the tech-               Migration of pervasive computing technol-          in” to the chassis structure without requiring
nologies we develop independently of the               ogy from the lab to the home will require sys-        skilled labor. Finally, interior finishes are
people using them. We need to study people             tems that provide value to homeowners without         snapped on to cover joints and wiring raceways.
using the technology in realistic, nonlaboratory       unreasonable cost and disruption to their liv-        The resulting structure will be easier to change
settings for long periods of time and then mea-        ing environments. For example, technologies           than conventional housing, require less expen-
sure whether our interventions lead to learning        such as wall-sized displays will be slow to           sive labor during construction, allow more
and behavior change.                                   migrate to the home if they require skilled labor     money to be spent on higher-quality materials
   To address this need, we have designed a full-      on site to conduct disruptive procedures such         and technologies, and easily accommodate sens-
scale single-family home with an integrated and        as tearing out drywall or installing digital and      ing infrastructure and new output technologies.
ubiquitous sensor architecture. This facility will     electrical cables.                                       The goal of the system is to empower home-

APRIL–JUNE 2002                                                                                                                  PERVASIVE computing              83
I N T E G R AT E D E N V I R O N M E N T S

                                                                                                                 like to see all the video and audio data of activ-
                        Camera
                                                                                                                 ity in rooms where there are television displays
                                                                                                                 and the television was on.” The researcher
                                                                                                                 could then manually view only the small subset
                                                                                                                 of relevant data using a video retrieval tool.12
                                                                                                                 Other types of data will be available as well,
                                                                                                                 such as positions of people in the environment.
                                                                                                                 An algorithm developer could use the living lab
                                                                                                                 to acquire probabilistic data about people’s
                                                                                                                 movement around environments throughout
                                                                                                                 typical days. The lab will provide an excellent
                                                                                                                 resource with which to study, in ways not pre-
                                                                                                                 viously possible, how certain technologies dis-
                                                                                                                 rupt activity in the home.5
                                                                                                                     For instance, we have built a prototype of a
                                                                                                                 new research tool for a living lab with ubiqui-
       (a)                                                 (b)
                                                                                                                 tous sensing. The tool queries users for informa-
     Figure 2. (a) Sensor packs in the beams of the living lab chassis structure will contain tiny               tion of value to researchers using a technique we
     cameras that can periodically snap images of the activity in the environment. (b) Image-                    call image-based experience sampling and reflec-
     based experience sampling and reflection algorithms will use a combination of ubiquitous                    tion.13 Figure 2 shows how it exploits the sensor
     sensing, context awareness, and user interfaces for portable computing technologies to                      infrastructure. Suppose a researcher wishes to
     acquire information from home occupants about how they perceive their environment.                          collect data on the types of activities that a per-
                                                                                                                 son is engaged in. As the occupants in the living
                                                                                                                 lab go about their business, the system samples
     owners so that they can replace interior and exte-     infrastructure to both enable and study new          and stores images without disrupting the occu-
     rior infill panels at will without costly and messy    context-aware applications. We are currently         pants’ activity. Then later, at their convenience
     custom work. Infill components might include           creating visually based, real-time people- and       (for instance, when commuting to work on a bus
     integrated wall–floor assemblies, specialty mill-      object-tracking algorithms for environments          or waiting in a long line), they can view the
     work with transformable elements, display sys-         that use these sensor packs.                         images on a portable computing device and eas-
     tems, networked appliances and devices, and so                                                              ily enter information about what they were doing
     on. The House_n research team is establishing          Studying physical–digital interactions               or how they were feeling about themselves or
     criteria that might inform industry standards.         in the home                                          their environment. The rich contextual infor-
        The team is designing the chassis compo-                Relatively little research has been done on      mation provided by the image or video clip trig-
     nents to incorporate multipurpose “sensor              the relationship between the home and tech-          gers the occupant’s memory of the moment when
     packs.” These are standardized sets of sensors         nology, given the importance of the home in          the sample was taken. Researchers can use the
     that easily plug into the chassis beams at regu-       life.11 Once we have constructed a living labo-      data collected directly or through algorithms that
     lar intervals (one every six to eight feet on aver-    ratory, we will use ubiquitous sensing to quan-      respond to user preferences.
     age) and can provide computing applications            titatively and qualitatively study human behav-          Finally, we are studying the relationships
     with access to data on interior and exterior envi-     ior in the home—with and without new                 between the environment and the use of digital
     ronmental conditions to support context-aware          technologies. To do so, we are developing tools      information. For example, by experimenting
     computing. The beams are nine feet above the           that will use the living lab’s sensing infrastruc-   with displays integrated into devices, we found
     floor. Each pack will minimally consist of a           ture to acquire and semiautomatically annotate       that physical constraints affect digital design.
     fixed, wide-angle color camera, a microphone,          data of interest to researchers.                     Figure 3 shows a scenario where a digital table
     and temperature sensors, but other sensors can             The lab’s infrastructure will make it possible   and a digital counter are in an environment that
     also be included (for example, an infrared cam-        to continuously acquire video, audio, and appli-     uses computer vision sensing to detect the table’s
     era or a particulate sensor). We are developing        ance-level data from every part of the environ-      position. From developing such mock-ups, we
     the sensor packs so that as costs drop they can        ment. Researchers will be able to identify the       noticed that the position and surface properties
     be retrofitted into existing environments with         types of situations they would like to study and     of physical materials in the space implicitly con-
     only a small amount of labor.                          then have access to a variety of data on that        vey meaning about their functionality to users.
        Using this construction methodology, we will        activity. For instance, a researcher might be        People generally expect surfaces with the same
     build a laboratory facility that has visual and        studying television viewing in the home. We are      appearance to have the same digital properties.
     auditory input sensors at regular intervals in         considering the development of tools that will       The user’s interaction model is established by
     every room. Researchers will be able to use this       let that researcher make a request such as, “I’d     seeing one interaction example on a surface and

84   PERVASIVE computing                                                                                                         http://computer.org/pervasive
(a)                                                         (b)                                              (c)

Figure 3. A digital table and counter in our lab were (a) designed for edge-to-edge video projection so that they can be abutted
against each other. Visual tracking (b) can be used to estimate objects’ positions and automatically create larger display surfaces
(c) when objects are in close physical proximity.

propagating that model along surfaces with the          and behavior change. Studies run in this lab will      new evidence to the diagnosis information. If
same “architectural” properties. For instance,          have a limited sample size (that is, one house         the system detects progression toward CHF
users who observe data projected onto one part          and a small set of long-term occupants), and           onset, it can notify the person that a medical
of a surface or who observe touch sensitivity on        must address experimental problems such as             professional should be contacted.
one part of a surface assume those properties           the Hawthorne effect (a distortion of research            Systems that provide information to people
extend to all parts of that architectural compo-        results caused by the subjects’ response to the        via computerized telephone conversations have
nent with similar physical properties.                  special attention they receive from researchers).      proven effective at motivating behavior change.16
   Similarly, users believe that physical expecta-      However, our discussions with researchers in           Will systems that present motivational informa-
tions should be matched in the digital world. A         fields as disparate as preventive medicine and         tion and acquire data for preventive diagnosis
kitchen table should move around, and com-              product development have led us to believe the         such as our CHF system work when placed in
bining two tables should be possible—combin-            living lab will enable studies that can take place     the complex environment of the home? Why or
ing the functionality involves no more than             in no other way.                                       why not and to what degree? How does context
putting the physical objects together. An envi-             For instance, we have built a prototype sys-       affect the way the information is received and
ronment that can automatically track movable            tem that uses a PDA device and context recog-          attended to? This is the type of investigation we,
objects such as people and tables and that has          nition (in this case location within the environ-      along with our collaborators, foresee occurring
“edge-to-edge” digital devices can extend the           ment and proximity to large objects) to detect         in the living lab.
physical metaphor to the digital domain, as in          the onset of congestive heart failure.15 The soft-        Different types of studies will require a vari-
the figure. When physical devices merge, digital        ware uses a Bayesian framework not only to             ety of research protocols. In addition, exactly
interfaces should merge as well.14 Researchers          integrate evidence of heart failure but to select      who lives in the home and for how long will be
will be able to study these issues in the living lab.   meaningful questions to ask a person in a home         determined based on the studies that are
                                                        given the context. Cameras monitor the envi-           selected for the home after construction. As an
Measuring effectiveness                                 ronment and detect contextual cues (for exam-          example of how measurement of learning might
   The last function of the living lab is to enable     ple, if someone is probably sleeping because of        take place, consider the earlier cross-breeze
the evaluation of certain types of pervasive            lack of movement and proximity to the bed). A          example. Assume that the set of studies selected
computing applications. We are particularly             diagnostic system pools evidence acquired over         for the home (there will be nonconflicting stud-
interested in studying how context affects the          the last month and, at any moment, can deter-          ies running in parallel) require three-week stays
presentation and motivational impact of infor-          mine which question that is appropriate for the        of subjects. During each subject’s stay, the win-
mation presented in the home environment                given context will yield the most valuable evi-        dow could be monitored for open and close
over long periods of time.                              dence. The home occupant carries a PDA                 events using sensors embedded in the window
   Our group is collaborating with researchers          device. Whenever the person pulls it out to use        or visual sensors in the chassis beams. Prior to
in a variety of fields as we design the living lab.     it, the PDA displays a simple but meaningful           entering the home, subjects would complete a
The challenge is to design a facility and infra-        question given the user’s current context. The         survey that assesses their understanding of their
structure that lead to verifiable and quantifi-         person quickly clicks one multiple-choice              living environment, including climate control
able advances in understanding how to use per-          answer with almost no interruption to the              and cross-breeze management. While each set
vasive computing in homes to motivate learning          intended task. Meanwhile, the system adds this         of subjects is living in the home, they will

APRIL–JUNE 2002                                                                                                                  PERVASIVE computing                85
I N T E G R AT E D E N V I R O N M E N T S

     manipulate the windows. If need be, the system        We invite researchers who might be interested                 E. Bergman, ed., Morgan Kaufmann, San Fran-
     could also manipulate the temperature in the          in conducting studies in a living laboratory to               cisco, 2000, pp. 335–360.
     home to ensure that the occupants open and            contact us.                                               10. C. Narayanaswami and M.T. Raghunath,
                                                                                                                         “Application Design for a Smart Watch with a
     shut the windows. The system will log the                                                                           High Resolution Display,” Proc. 4th Int’l Symp.
     video, audio, and other (for example, temper-                                                                       Wearable Computers (ISWC’00), IEEE CS Press,
     ature) sensor data for any instance where some-                                                                     Los Alamitos, Calif., 2000, pp. 7–14.
     one manipulates a window.                             ACKNOWLEDGMENTS                                           11. D. Hindus, “The Importance of Homes in Tech-
         The sample size would be small (n < 24 for                                                                      nology Research,” Proc. 2nd Int’l Workshop
                                                           Kent Larson directs the House_n: MIT Home of the              Cooperative Buildings (CoBuild 99), Lecture
     eight sets of subjects where three members of         Future Consortium and leads the research on the chas-         Notes in Computer Science 1670, Springer-Ver-
     each family can operate the windows). Although        sis–infill system. The ideas presented here benefited         lag, Berlin, 1999, pp. 199–207.
                                                           greatly from discussions with Chuck Kukla, Dan Carlin,
     some studies might only show trends and not           Byron Stigge, Xiaoyi Ma, and the entire House_n team      12. P. Castro et al., “Managing Context for Internet
     statistically significant results, researchers will   of staff and students.                                        Videoconferences: The Multimedia Internet
                                                                                                                         Recorder and Archive,” Proc. Multimedia Com-
     be able to qualitatively study the technology’s                                                                     puting and Networking 2000, vol. 3969, SPIE,
     impact using the data automatically collected                                                                       Bellingham, Wash., 1999; http://godfather.cs.
     before, during, and after the point of decision.                                                                    ucla.edu/publications/pdf/MCN2000.pdf.
     Does the user appear to be attending to the infor-                                                              13. S. Intille, C. Kukla, and X. Ma, “Eliciting User
     mation at the point of decision? Does the pre-                                                                      Preferences Using Image-Based Experience Sam-
                                                                                                                         pling and Reflection,” Proc. Conf. Human Fac-
     sentation of the information create a disruption?     REFERENCES                                                    tors and Computing Systems, ACM Press, New
     The precise experimental protocols will differ         1. C. Kidd et al., “The Aware Home: A Living Lab-            York, 2002, pp. 738–739.
     based on the problems being studied. Our goal             oratory for Ubiquitous Computing Research,”           14. N Streitz et al., “i-LAND: An Interactive Land-
                                                               Proc. 2nd Int’l Workshop Cooperative Buildings            scape for Creativity and Innovation,” ACM Conf.
     is to design a tool that lets researchers from var-       (CoBuild 99), Lecture Notes in Computer Sci-              Human Factors in Computing Systems, ACM
     ious disciplines design and execute studies that          ence, vol. 1670, Springer-Verlag, Berlin, 1999,           Press, New York, 1999, pp. 120–127.
     cannot be accomplished without the home’s                 pp. 191–198; www.cc.gatech.edu/fce/house/
                                                               cobuild99_final.html.                                 15. S.S. Intille, K. Larson, and C. Kukla, “Just-In-
     ubiquitous sensing infrastructure.                                                                                  Time Context-Sensitive Questioning for Preven-
                                                            2. M. Mozer, “The Neural Network House: An                   tative Health Care,” to be published in Proc.
                                                               Environment that Adapts to Its Inhabitants,”              AAAI 2002 Workshop on Automation as Care-
                                                               Proc. AAAI Spring Symp. Intelligent Environ-              giver: The Role of Intelligent Technology in Elder
                                                               ments, tech. report SS-98-02, AAAI Press, Menlo           Care, AAAI Press, Menlo Park, Calif., 2002.

     O
                                                               Park, Calif., 1998, pp. 110–114.
                  ur proposal for a home that                                                                        16. R. Friedman, “Automated Telephone Conversa-
                  teaches occupants how to take             3. M. Coen, “Design Principles for Intelligent Envi-
                                                                                                                         tions to Assess Health Behavior and Deliver
                                                               ronments,” Proc. 15th Nat’l Conf. Artificial Intel-
                  control raises the following three                                                                     Behavioral Interventions,” J. Medical Systems,
                                                               ligence, AAAI Press, Menlo Park, Calif., 1998,
                                                                                                                         vol. 22, no. 2, Apr. 1998, pp. 95–102.
                  challenges, among others, that               pp. 547–554.
     would benefit from further exploration:                4. J. Rodin and E. Langer, “Long-Term Effects of a
                                                               Control-Relevant Intervention with the Institu-
                                                               tional Aged,” J. Personality and Social Psychol-
     • Measuring learning or behavior change.                  ogy, vol. 35, no. 12, Dec. 1977, pp. 897–902.
                                                                                                                     For more information on this or any other computing
                                                                                                                     topic, please visit our Digital Library at http://
       How can we develop algorithms and systems
                                                            5. W. Edwards and R. Grinter, “At Home with              computer.org/publications/dlib.
       that use context-aware sensing to measure               Ubiquitous Computing: Seven Challenges,” Proc.
       the impact of new technologies on learning              3rd Int’l Conf. Ubiquitous Computing, Lecture
       and behavior over long periods of use?                  Notes in Computer Science 2201, Springer-Ver-
                                                               lag, Berlin, 2001, pp. 256–272; www.parc.
     • Using context-based simulation. One of the              xerox.com/csl/members/grinter/ubicomp.pdf.                                        the    AUTHOR
       most effective ways to learn is through              6. C. Pinhanez, “The Everywhere Displays Projec-
                                                                                                                        Stephen S. Intille is a research scientist working
       guided exploration via simulation. Can we               tor: A Device to Create Ubiquitous Graphical
                                                                                                                        with the Changing Places/House_n: MIT Home of
       create real-time simulations of environments            Interfaces,” Proc. UbiComp 2001: Ubiquitous
                                                                                                                        the Future Project in the MIT School of Architec-
                                                               Computing, LLCS 2201, Springer-Verlag, Berlin,
       that can be used to show people the impact              2001, pp. 315–331; www.research.ibm.com/ed/
                                                                                                                        ture and Planning. His research focuses on the
                                                                                                                        development of context recognition algorithms
       of their actions at the point of decision? For          publications/ubicomp01.pdf.
                                                                                                                        and interface design strategies for ubiquitous com-
       instance, if I open this window now to this          7. C. Seligman, L. Becker, and J.M. Darley, “Behav-         puting environments. He is creating pervasive
       degree, what is the estimated impact on the             ioral Approaches to Residential Energy Conser-           technologies for persuasive environments that
                                                               vation,” Energy and Building, vol. 1, 1977–1978,         motivate behavior change over long periods of
       breeze in my home one hour from now?                    pp. 325–337.                                             time. A special interest is behavior change tech-
       Also, can these systems exploit the Internet                                                                     nologies for proactive, preventive health care in
                                                            8. R. Winkler and R. Winett, “Behavioral Inter-
       for automatically acquiring required data?              ventions in Resource Conservation: A Systems             the home. He received a PhD in media arts and
                                                                                                                        sciences from MIT, an SM from MIT, and a BSE in
     • Detecting the point of decision. How can we             Approach Based on Behavioral Economics,”
                                                                                                                        computer science and engineering from the Uni-
       identify the point of decision for various              American Psychologist, vol. 37, 1982, pp.
                                                                                                                        versity of Pennsylvania. Contact him at intille@
                                                               421–435.
       activities and then detect these moments in                                                                      mit.edu.
                                                            9. B. Fogg, “Persuasive Technologies and Netsmart
       time automatically?                                     Devices,” Information Appliances and Beyond,

86   PERVASIVE computing                                                                                                              http://computer.org/pervasive
You can also read