Intelligent
Vacuum Cleaner
Systems
www.servomagazine.com/index.php?/magazine/article/february2011_Blankenship
by John Blankenship and Samuel Mishal
Creating a simulation of a robotic vacuum cleaner makes it easy to explore more
advanced behaviors than those found in commercial products. Imagine your
robot vacuum automatically scurrying to clean up when you accidentally spill
some popcorn.
In the November ‘ 10 SERVO, Editor Bryan Bergeron challenged readers to find economically feasible solutions for making home vacuuming robots more intelligent. Current systems generally perform their work by randomly roaming around the room, hoping to accidentally encounter bits of litter. One
of Bryan’s goals was to have a system that could detect
accidents (such as spilled popcorn) and immediately move
to the affected region to clean the area quickly and
efficiently.
There are many aspects of a vacuum application that
have to be solved before one can actually be built. First,
some form of sensor is needed that can detect debris on
the floor and pinpoint the room coordinates of its location.
The first thing that comes to mind is a ceiling mounted web
cam, but even fragments the size of popcorn could be hard
to detect with a standard camera. Finding an appropriate
sensor might require thinking out of the box. Perhaps some
types of litter could be more easily detected with an IR or
UV camera and it is not inconceivable that someone might
come up with a totally new approach.
Even if a suitable sensor can be found or devised, there
are other problems that must be considered. Once the
robot knows where the dirt is, it must still navigate through
the room to the desired destination while avoiding objects
encountered along the way. The robot must also be able to
determine when its battery is low and find its way to a
charging station. Of course, people in the room should be
avoided and changes in the placement of furniture should
not affect the robot’s performance.
FIGURE 1.
SERVO 02.2011 37
