Roomba is a vacuum-cleaning
robot sold by iRobot that moves
around flat surfaces autonomous-
ly. It is programmed to optimally
vacuum a place, with sensors that
enable it to detect obstacles; for
instance, it does not fall down
stairs. It says a few sentences,
such as “Roomba error 2” and
“Please charge Roomba.” The
models used for our research all
have a docking station where the
Roomba goes back to recharge,
three buttons (start; dock, to go
back to the docking station; and
spot, to vacuum around a precise
spot), and some have virtual walls
(infrared walls that force the robot
to stay within a defined perimeter).
To use it, one can push a button
and then walk away. Roomba does
not scratch any furniture. It does
not break anything except very
fragile objects in its path. Never-
theless, it is likely to get caught up
in cables lying around.
was that they managed to reduce
their cleaning efforts with few adaptation efforts (little to be learned,
few modifications of their homes
and habits). For the other households, Roomba still needed some
design modifications. First, it was
not usable across a wide variety of
physical environments (e.g., houses
with clutter on the floor or with
thick carpets or door sills). Second,
it needed to be more convincing in
terms of reducing cleaning efforts
(e.g., vacuuming faster while being
less noisy) and increasing cleanliness (e.g., not releasing balls of
dust, vacuuming well in corners).
We believe that what could also
increase adoption is more transparent communication of Roomba’s
capacities (e.g., how long it normally
takes to vacuum how many square
meters). People are not interested in
knowing about Roomba’s features
but rather in how to optimally use
it—for instance, how to use its virtual walls or whether or not to leave
it to roam the house on its own.
robot that talks does, at first, raise
enthusiasm. On the other hand, one
of our participants said he did not
want to have conversations with his
vacuum cleaner. We believe that
in order to answer these questions,
an appropriate approach would be
to do a comparative study of new
generations of robots based on longitudinal ethnographic research.
Good things come to those who are
patient—and willing to give new
technologies a try.
1. This research is part of a wider nationally funded
research program which ambitions to foster the
introduction of robots in private environments
2. According to Sung et al. [ 5, 8], usage patterns of
a Roomba settle after two months.
3. Venkatesh, V., Bala, H. Technology acceptance
model 3 and a research agenda on interventions.
Decision Sciences 39, 2 (2008).
4. Brown, S., Venkatesh, V. Model of adoption of
technology in households: A baseline model test
and extension incorporating household lifecycle.
MIS quarterly 29, 3 (2005), 399-426.
5. Sung, J., Christensen, H.I., and Grinter, R.E.
Robots in the wild: Understanding long-term use.
Proc. HRI 2009.
6. Fink, J., Bauwens, V., Mubin, O., Kaplan, F., and
Dillenbourg, P. People’s perception of domestic
robots: Same household, same opinion? Proc.
ICSR 2011 (Amsterdam, The Netherlands), accepted for publication.
7. Forlizzi, J., DiSalvo, C. Service robots in the
domestic environment: A study of the Roomba vacuum in the home. Proc. HRI 2006 (Salt Lake City,
UT). ACM Press, New York, 2006, 258-265.
8. Sung, J., Grinter, R.E., and Christensen, H.I.
Domestic robot ecology. An initial framework to
unpack long-term acceptance of robots at home.
Int J Soc Robot 2, 4 (2010), 417-429.
March + April 2012
households? At this stage, we believe
one should talk about an evolution
rather than a revolution. First, the
domestic robots available today on
the market are still used in few situations (e.g., cleaning gutters, mowing lawns, etc.). Second, according to
our study, only a minority of households adopted Roomba. Indeed, it
did not prove robustly adaptable to
any kind of physical environment.
Third, Roomba has not had a major
impact on practices at home. Even
though previous research indicated
that Roomba would motivate people
to renegotiate task responsibilities
around cleaning [ 7, 8], we did not
notice that; it simply enabled some
participants to get rid of some vacuuming tasks.
Lessons for Roomba’s Designers
Roomba proved to be an outstanding cleaning companion for some of
our families. What convinced them
© 2012 ACM 1072-5220/12/03 $10.00
Lessons for Future Service Robots
How can we transfer these lessons
to future robots? Future domestic-service robots should definitely
focus on enabling their users to
quickly experience their usefulness,
with low adaptation efforts. When
it comes to solving the issue of the
physical space, there are different
paths to follow. Should Roomba
really become an all-around
vacuum cleaner, riding up stairs,
dealing with books and thick carpets? Wouldn’t a vacuum cleaner be
more efficient if it were specialized?
We have seen in our research that
the cleaning pain points are often
linked to specific spots in the house
(e.g., the kitchen table, the entry
hall). It also isn’t clear what should
be undertaken in terms of interactive capabilities. On the one hand, a ABOUT THE AUTHORS Valérie Bauwens works in the CRAF T department at the École Polytechnique Fédérale de Lausanne (EPFL) studying the adoption of robots in homes. She is the founder of www.human- centricity.com, which specializes in ethnographic market research and teaching ethnography.
Julia Fink is a Ph. D. candidate in the CRAF T department at the École Polytechnique Fédérale de Lausanne (EPFL), studying human-robot interaction in domestic environments. She has a multidisciplinary background in media, technology, and communication.