In modern poultry processing facilities, humans must still manually place carcasses
onto a moving cone line for further processing. Robotic handling systems have been
ineffective in this area mainly because of the speed required to perform the task
and the random orientation of the carcasses being handled. However, recent developments
in image processing have made it possible for robotic systems to operate in such
unstructured environments. Researchers at Georgia Tech have taken advantage of those
advances with the development of a prototype automated cone loader.
Using a research robot, the development team has constructed a laboratory-scale experimental
prototype. The prototype uses a commercial 3-D camera system, which provides the
position information needed to move the robot to the bird as well as identify the
orientation of the bird. A newly designed end effector grasps and manipulates the
irregular and non-rigid texture of the raw product.
In the processing plant, the deboning process begins with the placement of the front
half of the bird onto the moving cone line. Gary McMurray, project director and lead
developer of the automated cone loader, explains that for correct placement of the
bird onto the cone, the front half must be properly oriented, and the open cavity
has to be secured on the cone. The cone is a conical-shaped plastic fixture; the
rib cage of the bird slips over the cone. The cone is vertical when the bird is attached,
and the force of gravity is sufficient to hold the bird on the cone. In addition,
notes McMurray, the cone contains a keel pin that is designed to penetrate the fatty
pocket along the inside of the bird’s keelbone. Additional force is required for
the keel pin to penetrate the fatty pocket. The keel pin fixtures and prevents rotation
of the bird during successive deboning operations.
“Grasping such an object is a difficult task due to the slippery nature of the raw
product as is maintaining grasp stability on the bird as it is placed on the cone,”
says McMurray. “In addition, the variation of the raw product’s size, shape, weight,
and orientation had to be considered in the design.”
McMurray notes that the prototype system addresses this by employing specially designed
software that moves the cone loader’s end effector or hand to the bird, grasping
and then placing it onto the cone line. This is accomplished, he adds, regardless
of the bird’s location or orientation.
In recent performance tests, the grasping success rate was 100 percent. The cone-loading
rate (placement of the bird onto the keel pin) was 93.6 percent compared to a manual
success rate of 92 percent. According to McMurray, the major mode of failure for
the end effector was related to the inability of the system to have the keel pin
hit its intended target. Each carcass was grasped correctly and placed onto the cone;
however, the keel pin did not puncture the center of the fatty pocket. This premature
puncture of the carcass, he says, can be attributed to the variability in the carcass’
shape that the robot and end effector were unable to correct.
Nevertheless, McMurray says, the performance of the end effector demonstrated the
ability of the automated system to match the accuracy of the manual laborer.
“The implementation of this technology lays the groundwork for introducing flexible
automation in production tasks that were once considered impossible to automate,
such as the rehang of birds after the chiller and placing raw product into trays,”
adds McMurray.
