Last month, we saw how a robot could find and identify a ball and
goal, allowing it to engage in a simplified version of soccer. This
month, we will examine how the principles of identification used
with the soccer robot can allow a mobile robot to find and retrieve
objects with a simple robot arm and then move them to a desired
destination.
Robotic Sensors
Part 2: Their Use
With Robotic Arms
by John Blankenship and Samuel Mishal
www.servomagazine.com/index.php?/magazine/article/april2013_Blankenship
Discuss this article in the SERVO Magazine forums at http://forum.servomagazine.com
As mentioned, last month we used a simplified soccer-playing robot to demonstrate that a robot’s
ability to cope with its
environment can be improved by
giving it a rudimentary ability to
identify objects of importance. The
ball trap on the soccer robot was
essentially a non-motorized
gripper, and the robot itself could
be thought of as a robotic arm
with one degree of freedom.
Because of these similarities, it
makes sense to expand on
principles discussed last month by
applying them to a more capable
robotic arm.
Before we begin though, we need
to examine robotic arms themselves. If
you search the ’net for robotic arm
images, you will find hundreds of
arms that are suitable for the robot
46 SERVO 04.2013
hobbyist. In almost every case though,
the arms have a major flaw — the
arms and grippers have little — if any
— sensory capability. What’s worse is
that most arms have not even been
designed to allow sensors to be easily
added. This limits the arm’s activity to
pick-and-place maneuvers where the
objects to be manipulated are
assumed to be found at specific
positions, sometimes with specific
orientations. Industrial assembly lines
can often make such assumptions, but
a hobby arm — like most hobby robots
— needs sensors to properly interact
with its environment.
The lack of sensors may seem
acceptable at first, because most low
cost arms utilize servomotors to
power their joints, allowing the arm to
be positioned without the need for
sensory feedback. This is not
unreasonable as long as maximum
loads are not exceeded, plus it has
significant cost advantages. High-end
digital servomotors — such as
Dynamixels — have built-in sensory
capabilities that can provide
information about the position and
load associated with an arm joint.
However, this does not eliminate the
need for gripper-mounted sensors that
can be used to guide and control the
arm’s movements.