A Robot In Every Home?
At the bottom of
Figure 2 a surveillance
robot is depicted.
My first impression of
such a robot roaming
about the outside of
someone’s home is that
somebody is going to
steal that thing! Some
security! Why have a
mobile security robot?
Why not just augment a
nice monitored home
security system with
extra TV cameras? Leave
the high-tech mobile robots to warehouse and military
The “Simple” Task of
Folding a Towel
One type of robot that jumped out at me from
Figure 2 was the laundry folding robot. This task is not as
loathed as toilet cleaning, but a few people enjoy the task
of folding laundry. Last year, as Willow Garage’s PR- 2 made
the circuit of You Tube videos, one of the most interesting
was the $400K robot folding laundry. What many people
do not realize is that it took a lot of programming, plus the
actual folding operation was a lot longer than shown in the
videos. Willow Garage was in no way trying to sell this
amazing robot as a laundry room attendant. They were
simply illustrating how the robot’s vision system can work in
conjunction with the two articulated arms to perform a
Researchers at UC Berkeley developed a cloth-grasping
algorithm to be used by
their PR- 2 to allow it to
grasp, pick up, and fold
towels it hadn’t previously
analyzed. Figure 4 is a
photo of the experiment
that was written up in a
2010 ICRA paper and
accompanying video I
mentioned earlier. The
test involved 50 separate
towels and a pile of five.
The robot had an 81%
success rate in grasping
the towels, and took an
average of almost 25
minutes to fold each one.
The greatest amount of
time was required for the
PR- 2 to just determine
where to grasp each towel. (Can you imagine how difficult
it would be for this very sophisticated robot to grasp inside-out socks, various sizes and types of underwear, wrinkled
shirts, or lingerie and correctly fold them?)
FIGURE 3. Joe Engelberger's home robot.
Specialty Tasks Within the Home
FIGURE 4. Willow Garage's
PR- 2 towel-folding demo
at UC Berkeley.
FIGURE 5. The Qbo robot
from The Corpora.
I started out this article by commenting how on-the-money Gates was in his article and then managed to find
problems with each of his examples. All of the types of
robots that he mentioned are viable concepts and many are
high on the list of people’s needs for a robot in the home.
It seems that simple tasks for a human are complex tasks
for a robot.
If one adds up the unpleasant chores of toilet cleaning
( 20 minutes a week) and laundry operation for a single
person ( 10 minutes loading and starting the washing
machine; five minutes transferring loads from the washer to
dryer, and 25 minutes removing, folding, and stowing), you
have an hour of menial human labor costs. Add to that the
time to do light housecleaning (no Roombas here) and
some kitchen prep work, and you might have two to three
hours of a human’s time. Once a month or so, oven and
fridge cleaning can be included, as well as heavier spring
cleaning tasks. A lawn service is certainly preferable to a
$1,000 lawn robot that is used an hour a week, as is a
basic home security system over a mobile robot.
Food and medicine dispensing robots get even more
complicated. If the only resident of the home is totally
bedridden, then any foreseeable home robot design will
have a tough time being accepted by the FDA. The FDA
Code of Federal Regulations, part 890.5050, describes a
“Daily Activity Assist Device” that might be applicable to
most personal assistant robot designs, but regulations
are very strict for any ‘medical’ device that comes in
contact with a human. So, how exactly can we make
78 SERVO 12.2011