Académique Documents
Professionnel Documents
Culture Documents
fe ature
Julia Fink
École Polytechnique Fédérale de Lausanne | julia.fink@epfl.ch
Domestic robots have slowly found holds and observed them over a
their way into some of our homes period of six months [2].
and onto the shelves of major stores We recruited households with and
selling technical appliances. Who without children, pets, and gardens.
hasn’t already seen or heard of We analyzed cleaning habits before
robots that vacuum or mow the Roomba. We then observed how
lawn? As researchers in robotics, they evolved from the moment we
we feel this growing commercial brought them the robot: at installa-
success is a great opportunity to tion, after two weeks, and then two
learn about robot adoption pro- and four months after installation.
cesses. Leaving the marketing buzz
and usual fantasies about robot Adoption Factors
invasions aside, we are curious to Two weeks after we brought Roomba
find out how robots are perceived to their homes, two households
by users. Are robots revolution- were on the verge of stopping
izing people’s practices at home? using the robot (we called them
Understanding the adoption of “rejectors”), four were sporadi-
such robots is also central, as it cally using it (the “skepticals”), and
helps to pinpoint crucial factors only three were completely con-
to be taken into account while vinced about it (the “aficionados“).
designing new robots. Other ques- After two months this classifica-
tions we wish to consider include: tion was still valid and remained
What convinces people to adopt so until the end of our trial.
i n t e r a c t i o n s M a r c h + A p r i l 2 0 12
61
fe ature
3. Practical relation to I don’t see what I have to learn I have learned how to use the
technology about Roomba. technology optimally.
5. Emotional relation to Few personal interactions with Roomba. It has become less
technology fun over time.
•F
igure 1. Key
adoption criteria 7. Economic utility What about costs of future spare parts?
per user groups.
one of the most important factors before or cleaner, with less effort. In ties climbing onto some rugs and
affecting adoption. Interestingly, the contrast, Roomba had not managed carpets and tends to release some
identified criteria match quite well to solve any major pain point for the dust when doing so. One of our
with those identified in other stud- skepticals or the rejectors. A rejec- participants had a nice rug in her
ies on adoption of technology [3,4]. tor said, “I clean faster and better living room. Between her rug and a
Factor 1: Practical utility. People than Roomba. I lose time when I’m vacuum-cleaning robot that releases
agreed to participate in the study using Roomba.” dust, the choice was easy to make.
primarily because they thought Factor 2: Physical space. All of our Factor 3: Practical relationship to
they could spend less time vacu- Roomba aficionados either had an technology. Learning how to use
uming. Thanks to Roomba after optimal environment for Roomba Roomba was also an important
two months, all three households to work or adapted their physical aspect in adoption. All three aficio-
that adopted the robot had man- space to the robot to make it to nados had taken the time to learn
aged to get rid of some cleaning work optimally. This phenomenon how to use Roomba in an optimal
tasks. One mother eliminated is called “Roombarization” [5]. A way. One of the sporadic users might
her three-times-daily use of the single man spent a whole Sunday have optimized its use by making
broom to clean under the kitchen afternoon securing his Wi-Fi access use of the infrared barriers, which
table. Another mother managed to point to the wall in his living room would limit the areas in which
use her traditional vacuum only so that Roomba would not get Roomba worked. Unfortunately, she
once a week instead of three to tangled up in cables on the floor. was not willing to learn how to use
i n t e r a c t i o n s M a r c h + A p r i l 2 0 12
four times a week. A single man He also rearranged all the furni- them. Another aspect mentioned by
completely stopped using the tra- ture in his living room. Roomba three of our participants is the con-
ditional vacuum. For him, Roomba presented an opportunity to clean trol of the robot. A rejector said, “We
cleans well enough and solves all up his “cable salad” on the floor. don’t trust it fully,” complaining that
the other inconveniences related Interestingly, when he uses Roomba Roomba appears to move around in
to vacuuming (e.g., previously he he does not mind placing his coffee an uncoordinated way. The three
had tied the tubes of his traditional table on his couch to let Roomba aficionados had learned to trust and
vacuum cleaner with a rope in order vacuum optimally (see Figure 2). cope with Roomba’s surprising way
to store it neatly). In other words, None of the skepticals or rejec- of moving around.
these three households found that tors had done anything similar. Factor 4: Habits. All three aficio-
Roomba enabled a house as clean as For instance, Roomba has difficul- nados have adapted their habits or
62
fe ature
communicating with technology. In time—“Go and do your work in try it out for a few days. Cost of
terms of hedonism, nearly all of our the kitchen…”—but we could not replacement of spare parts and the
participants manifested enthusiasm observe lasting forms of emotional longevity of the battery were also
when we brought them Roomba, attachment. As with most of our often mentioned as potential issues
especially children (e.g., one wel- participants, a nine-year-old boy to consider when buying Roomba or
comed us wearing a T-shirt with said, shrugging his shoulders, “Well, even using it further.
a robot on it). In the first weeks of it is just a vacuum cleaner.”
adoption, Roomba generated enthu- Factor 6: Social influence. Many Not a Revolution, but an Evolution
siasm because of its erratic and participants showed or even lent Let’s come back to the initial ques-
surprising movements. One adult out Roomba to friends or colleagues, tion: Are these vacuum-cleaning
liked how Roomba would drive back which implies that Roomba played robots starting a revolution in our
63
fe ature
was that they managed to reduce robot that talks does, at first, raise
An Introduction their cleaning efforts with few adap- enthusiasm. On the other hand, one
to Roomba tation efforts (little to be learned, of our participants said he did not
Roomba is a vacuum-cleaning few modifications of their homes want to have conversations with his
robot sold by iRobot that moves and habits). For the other house- vacuum cleaner. We believe that
around flat surfaces autonomous- holds, Roomba still needed some in order to answer these questions,
ly. It is programmed to optimally
vacuum a place, with sensors that
design modifications. First, it was an appropriate approach would be
enable it to detect obstacles; for not usable across a wide variety of to do a comparative study of new
instance, it does not fall down physical environments (e.g., houses generations of robots based on lon-
stairs. It says a few sentences,
with clutter on the floor or with gitudinal ethnographic research.
such as “Roomba error 2” and
“Please charge Roomba.” The thick carpets or door sills). Second, Good things come to those who are
models used for our research all it needed to be more convincing in patient—and willing to give new
have a docking station where the terms of reducing cleaning efforts technologies a try.
Roomba goes back to recharge,
(e.g., vacuuming faster while being
three buttons (start; dock, to go
EndNotes:
back to the docking station; and less noisy) and increasing cleanli-
1. This research is part of a wider nationally funded
spot, to vacuum around a precise ness (e.g., not releasing balls of research program which ambitions to foster the
spot), and some have virtual walls dust, vacuuming well in corners). introduction of robots in private environments
(infrared walls that force the robot (NCCR http://www.nccr-robotics.ch/).
We believe that what could also
to stay within a defined perimeter). 2. According to Sung et al. [5,8], usage patterns of
To use it, one can push a button increase adoption is more transpar- a Roomba settle after two months.
and then walk away. Roomba does ent communication of Roomba’s 3. Venkatesh, V., Bala, H. Technology acceptance
not scratch any furniture. It does capacities (e.g., how long it normally model 3 and a research agenda on interventions.
not break anything except very Decision Sciences 39, 2 (2008).
fragile objects in its path. Never-
takes to vacuum how many square 4. Brown, S., Venkatesh, V. Model of adoption of
theless, it is likely to get caught up meters). People are not interested in technology in households: A baseline model test
in cables lying around. and extension incorporating household lifecycle.
knowing about Roomba’s features MIS quarterly 29, 3 (2005), 399-426.
but rather in how to optimally use 5. Sung, J., Christensen, H.I., and Grinter, R.E.
it—for instance, how to use its vir- Robots in the wild: Understanding long-term use.
Proc. HRI 2009.
households? At this stage, we believe tual walls or whether or not to leave
6. Fink, J., Bauwens, V., Mubin, O., Kaplan, F., and
one should talk about an evolution it to roam the house on its own. Dillenbourg, P. People’s perception of domestic
rather than a revolution. First, the robots: Same household, same opinion? Proc.
ICSR 2011 (Amsterdam, The Netherlands), accept-
domestic robots available today on Lessons for Future Service Robots ed for publication.
the market are still used in few situ- How can we transfer these lessons 7. Forlizzi, J., DiSalvo, C. Service robots in the
domestic environment: A study of the Roomba vac-
ations (e.g., cleaning gutters, mow- to future robots? Future domestic- uum in the home. Proc. HRI 2006 (Salt Lake City,
ing lawns, etc.). Second, according to service robots should definitely UT). ACM Press, New York, 2006, 258-265.
our study, only a minority of house- focus on enabling their users to 8. Sung, J., Grinter, R.E., and Christensen, H.I.
Domestic robot ecology. An initial framework to
holds adopted Roomba. Indeed, it quickly experience their usefulness, unpack long-term acceptance of robots at home.
did not prove robustly adaptable to with low adaptation efforts. When Int J Soc Robot 2, 4 (2010), 417-429.
that Roomba would motivate people vacuum cleaner, riding up stairs, adoption of robots in homes. She
is the founder of www.human-
to renegotiate task responsibilities dealing with books and thick car- centricity.com, which specializes in ethnographic
around cleaning [7, 8], we did not pets? Wouldn’t a vacuum cleaner be market research and teaching ethnography.
notice that; it simply enabled some more efficient if it were specialized? Julia Fink is a Ph.D. candidate in
the CRAFT department at the
participants to get rid of some vacu- We have seen in our research that
École Polytechnique Fédérale de
uming tasks. the cleaning pain points are often Lausanne (EPFL), studying
human-robot interaction in
linked to specific spots in the house domestic environments. She has
Lessons for Roomba’s Designers (e.g., the kitchen table, the entry a multidisciplinary background in
media, technology, and communication.
Roomba proved to be an outstand- hall). It also isn’t clear what should
ing cleaning companion for some of be undertaken in terms of interac-
DOI: 10.1145/2090150.2090165
our families. What convinced them tive capabilities. On the one hand, a © 2012 ACM 1072-5220/12/03 $10.00
64