Researchers at the Georgia Tech Research Institute (GTRI) recently completed the second year of a three-year project to rebuild their Intelligent Cutting and Deboning System, with the goal of making it into a bionic poultry deboner that can rival the cutting skills of the most experienced human.

The system uses machine vision and robotics to sever the tendons and joints on bird front-halves in preparation for the removal of the wings and breast meat, known as a butterfly cut. So far, the bird front-halves have been placed on stationary cones. The objective is to make cuts that maximize yield while eliminating bone chips in the cut meat.

Because it combines advanced robotics, image processing and statistical modeling in one device, the system can think and react to its environment just like a human — only it can do it better and faster. This makes it unique compared to other fixed automation systems that are currently on the market.

“The ongoing challenge with automating the deboning task has been creating a system that can handle the natural variability of poultry products. What makes our system groundbreaking is its flexibility and ability to customize cutting motions regardless of the bird’s size,” explains Dr. Ai-Ping Hu, GTRI senior research engineer and project director.

The so-called Intelligent Cutting 2.0 System has substantial hardware and software enhancements compared with its predecessor. The most significant advancement is its move to a 6-degrees-of-freedom knife robot. Unlike the previous 2-degrees-of-freedom system, 6 degrees better approximates a human’s wrist and hand (a human arm has a total of 7 degrees of freedom). This increased dexterity allows researchers to experiment with a wider range of cutting motions and geometries to arrive at optimal cuts.

In addition, newly developed advanced image processing algorithms are an order of magnitude better than previous methods. According to Hu, the system is now achieving millimeter-level accuracy in predicting the bird’s internal structure. This increased robustness allows the robot to define correlations between the bird’s external features and its internal joint structure — essential to making a clean cut with maximum yield and without cutting into bone.

Lastly, a newly written graphical user interface, or software wizard, streamlines the various processes of a bird cut, while data analysis of cutting motions made by human deboners helps to determine the optimum knife paths needed to make precision cuts. Interestingly, analysis of motion-capture data has shown areas of the bird that the human is not as efficient at cutting as a robot.

Hu says this last finding is particularly exciting to researchers as it is an indication that the robot will indeed maximize the amount of meat removed from the carcass, thus increasing yield. Yield means a lot to the poultry industry: higher yield equals higher profits.

“Because a robot’s motion is agile, faster and more precise, there are certain portions of the front-half cut that a robot can perform that can possibly result in a half to 1 percentage point more yield per bird,” says Hu.

To date, the Intelligent Cutting 2.0 System’s yield performance rivals that achieved by a human deboner, as demonstrated by quantitative yield comparisons conducted between the two. Hu says the next major milestones are to increase the robot’s cutting speed and to transition to a moving cone line operating at speeds found in a typical poultry processing plant. The team will also pursue industry partnerships that can help push near-term commercialization.

“Our goal in the remaining year of the project is to enhance our prototype machine to more closely resemble plant-ready equipment. We are also considering how existing commercial deboning equipment can be retrofitted with our technology,” Hu says. 

This article is an updated version of a previously published story from PoultryTech – to view the publication, visit www.atrp.gatech.edu.