Watt Poultry USA - March 2001
Not Your Typical Robot
Researchers are working on robotics for poultry processing that
may be able to adjust its movement based on what it sees, thereby mimicking
human flexibility
By Angela Colar
|
Georgia Tech researcher Gary McMurray
observes the prototype conveyor feed system and
IIBM case packer in laboratory setting. The system
will eventually be combined with a commercial weigh/price/labeler
in a poultry plant setting.
|
A robotics program begun 10 years ago in a research and
development laboratory at the Georgia Tech Research Institute in Atlanta
set out to design anything but your traditional robot. The aim was
to design a robotic system for poultry processing applications. It
would require capabilities well beyond the traditional robots that
have performed well in automotive and electronic manufacturing applications.
Those applications demanded extreme precision where environmental conditions
are moderate. Poultry processing applications are very different. As
a result, the Intelligent Integrated Belt Manipulator (IIBM) was conceived,
a robotic system that is able to tolerate poultry processing plant
environments, can be configured for a variety of poultry processing
applications and can handle products of different sizes and textures
with an accuracy similar to that of a human.
One of the first applications of the IIBM was in case-packing operations.
In a 1998 field trial at ConAgra’s poultry processing plant in
Gainesville, Ga., the IIBM accurately grasped tray-packed products
from a moving conveyor belt and placed them in packaging cartons for
shipping. The major success of the trial was the IIBM’s flexibility
in handling the different shape contours of tray packs. The system
performed best on the No. 8 tray packs. These tray packs were 10.5
by 7.5 inches in size and weighed between 3 and 5 pounds. The grasping
accuracy was about 100 percent at a 3-second cycle time. This result
proved that the IIBM’s gripper design is robust enough to secure
different types of products and product contours. The ConAgra test,
however, involved placing the IIBM in an existing packing cell designed
for a human packer. Researchers believe this cell impacted system performance
and are now focused on putting the IIBM in a more robotic-friendly
integrated cell that is part of a commercial weigh/price/label line
planned for Gold Kist’s Live Oak, Fla., poultry plant. The cell
includes custom-designed box- and tray- feed conveyors fabricated by
Peterson Engineering Service, Inc. of Gainesville, Ga.
|
The IIBM (Intelligent Integrated Belt Manipulator).
|
The IIBM is being programmed to have coordinated control over the throughput
of the entire integrated cell. Nominal throughput of the new cell in
laboratory trials has been 30 trays per minute. During the ConAgra
field trials, the IIBM packed trays at about 20 trays per minute. Wiley
Holcombe, project director, says the team believes this throughput
can be increased to about 45 trays per minute, which more closely matches
monthly average throughputs on a manual line.
The first steps taken on performance improvement were focused on
the IIBM’s gripper. Holcombe believes the reason the IIBM was
only able to run at 20 trays per minute during the ConAgra trial
was because
of grasping reliability. To address this challenge, the team designed
and built a centrifuge.
"
Vacuum cups of different sizes and materials were tested using the
centrifuge to stimulate the magnitude and direction of acceleration
that would be experienced on an IIBM with higher performance capability," explains
Holcombe. "The centrifuge allowed us to test the grippers at
much higher accelerations than we could generate on the case packer.
Based
on the results of recent tests, we are confident that we have a revised
gripper design that can grasp reliably at higher packing rates."
Researchers also conducted tests to establish what the best achievable
performance is with the IIBM’s existing hardware. They discovered
several of the drives were undersized for continuous performance
at higher speeds. Their plans are to replace these drives with units
capable
of meeting the higher performance needs. Once the system is upgraded
and its performance confirmed, it will be moved to Gold Kist for
field trials.
As for the trial setup, the team plans to position the case packer
adjacent to the weigh/price/label line. Cases will flow down the conveyor
to the packing location. Trays flow across a scale conveyor, under
a label applicator, and down a second conveyor to the pick location.
The tray conveyor surface is 26 inches above the case conveyor. A case
lifter raises the case 9 inches to reduce the required vertical stroke.
The case packer picks up the tray with a vacuum gripper, moves horizontally
in the direction of the conveyor flow, then moves down into the case
and releases the tray. It has an offset motion and a rotation to allow
it to pack a variety of packing patterns.
"
After the prototype has been running in the plant for several months,
we will hold a performance review at the plant. At that time, we will
identify any remaining problems that need to be corrected on the machine.
We will then address those problems and push the product toward commercialization," says
Holcombe.
Holcombe believes commercialization of the case packer is very promising.
A number of manufacturers expressed interest in the prototype during
a demonstration at the International Poultry Exposition held this past
January in Atlanta. It is also interesting to note that approximately
25 percent of the more than 250 poultry slaughter plants in the United
States use some sort of tray packing. And pre-packs represent 16.8
percent of the 27 billion pounds of total poultry production.
"
We estimate the total potential case packer sales to be 160 to 170
units in the poultry industry alone. There are also potential applications
in the red meat industry for case packing product in vacuum-sealed
plastic bags or in styrofoam trays," comments Holcombe.
Researchers have also been busy developing a new robotic control
program that incorporates vision directly into the robot’s
controller. This provides an almost humanlike flexibility to the
robot through
the addition of hand-eye coordination. It also allows the robot to
operate in a totally unstructured environment as opposed to the typical
structured industrial environment.
Most robotic systems in use today, including the IIBM, cannot adjust
to shifting conditions; for example, they cannot recognize that an
object has changed in size or weight or in orientation. Such adaptability,
however, could prove invaluable in a poultry processing environment,
where products vary in size, texture, and contour. Visual servo control
technology uses vision information as feedback to control the position
of the robot’s hand with respect to a particular target. As
a result, the robot is able to adjust its movement based on what
it sees,
thereby mimicking human flexibility.
This mimicking is accomplished through the use of novel software
control algorithms using commercial cameras and standard vision processing
techniques. The vision system features a digital camera mounted within
the workspace. In combination with sophisticated computer algorithms,
the unit can continually relay information to the robot’s guidance
computer, updating important information such as the speed and the
distance of objects to be handled. These algorithms are able to guide
the robot to pick up objects that are stationary or moving at relatively
high rates of speed. For pick-and-place operations, this ability
to react to its environment is highly desirable.
These visual servoing algorithms have been tested in simulation using
robotic systems ranging from a simple 2-degree of freedom to the more
common 6-degree of freedom industrial manipulators. These tests have
successfully demonstrated the validity of the algorithms.
The benefit of this technology is that it allows a robot to dynamically
react to its environment without the need for extensive modeling of
the robot, camera, or the environment itself. Researchers believe that
visual servo control has the potential to provide a low-cost, low-maintenance
automation solution for unstructured applications and environments
because of its ability to provide the same type of hand-eye coordination
that the human depends on so heavily. |
