A Robot That's Sharp
as a Knife

View larger

Failures at the Thanksgiving Day table notwithstanding, humans do a pretty good job of cutting up a bird.

So good, in fact, that the poultry industry has yet to successfully replace the human carver with a machine. Twice they've tried. In the 1990s and the 2000s, the industry made massive attempts to automate, with new de-boning equipment cropping up on poultry farms nationwide. The result was that a lot of robots got the heave-ho, as it quickly became apparent that humans remained superior at cutting meat from a bone.

"Yield is what drives the industry," says Gary McMurray, chief of the Georgia Tech Research Institute Food Processing Technology Division.

So when McMurray decided to take the plunge into automated bird carving, the first thing he looked at was the human. "We were very curious as to how the user was selecting the trajectories that they were making," McMurray says of the folks responsible for the boneless chicken meat found in supermarkets everywhere. "We noticed the human would grab the wing and just from looking at the bird…they kind of generally knew where the bones were. We were always confused as to how they knew that. In their minds they had some general idea about how to move the blade, about what they were doing before they did the cut. Then they changed based on feedback—that's what intrigued us."

Gary McMurray, chief of GTRI's Food Processing Technology Dvision, poses with the Intelligent Cutting and Deboning System. Photo: Gary Meek

Knife-Wielding Robot Arm

McMurray began to design a system that was capable of competing with humans both at sizing up a bird to determine the location of bones and at changing blade trajectory as it moved through different body parts. A chicken is "not a homogenous material," says McMurray. There's fiber running in various directions: water, fat, and, of course, tendons, ligaments, and bone. "All these real-world problems that make every bird unique. You begin to appreciate how amazing the human is. They can factor out these disturbances, and are able to do the job. When we started into this, we were certainly not expecting that much complexity."

In McMurray's system, three key points on each bird tell his machine where the bones are likely to be. Then a knife-wielding robot arm makes its first plunge. But as it proceeds, it recognizes the various substances it cuts through based on their "force signatures." The robot knows whether its blade is up against meat, tendon, or bone, and is able to change course accordingly.

Cost and Labor Savings

The result is the holy fowl of the industry: more meat off the bone (and without the risk of a bone chip). A typical plant will have seven deboning lines, with meat being harvested on both the left and the right. "In rough, back-of-the-envelope numbers, if you put in 100,000-dollar robots on each of those, it'll cost 1.4 million, maybe a bit more to put in," says McMurray. "Our goal is to save about two percent—that's five million in savings. Even at just a one percent, it pays for itself in six months, and that's not even counting in the labor savings,"

Once McMurray's efficient—and sensitive—carver proves its mettle on the chicken farm, it may have surprising applications elsewhere. In laparoscopic surgery, for instance, surgeons could know what organ a scalpel is pushing up against based on the feedback. "Having force feedback when you go to make a cut would be a fail-safe device. It would be sensitive enough to say, 'Now I'm cutting the skin, now I'm cutting a vein or an artery, now I'm cutting bone."

But before the carver starts helping surgeons, the system has got to hit the farms and debone a few million chickens. "It's not a finished system," says McMurray, "We're starting to understand the science of how to do this, and have some experimental results to justify our belief that we can do all this, but we're not ready to proclaim 'Mission accomplished.'"

Michael Abrams is an independent writer.

In their minds, [humans] had some general idea about how to move the blade, about what they were doing before they did the cut. Then they changed based on feedback—that's what intrigued us. Dr. Gary McMurray, Georgia Tech Research Institute Food Processing Technology Division

Tweet

Share

 Email

 Print