(Georgia Institute of Technology, 2009)
Hu, Ai-Ping; Grullon, Sergio; Zhou, Debao; Holmes, Jonathan; Holcombe, Wiley; Daley, Wayne; McMurray, Gary
Deboning operations are one of the largest users of on-line labor in today’s poultry plants. Efforts have been made over the years to automate this function, but to date have achieved only limited success. The main difficulty in this task is its unstructured nature due to the natural variability in the sizes of birds and their deformable bodies. To increase product safety and quality, the industry is looking to robotics to help solve these problems. This research has focused on developing a new method of automating the deboning of bird front halves. If this task can be automated, the technology would naturally be extended to other cuts and trimming operations in poultry and red meat. To accomplish this goal, the project team has been working for the past four years on the development of a sensor-based intelligent cutting system. This work is based on the development of a model for the cutting of bio-materials that can be extended to the cutting of meat, tendon, ligaments, and bone. When this model is combined with data from the tendon prediction system, the nominal cutting trajectory can be established and adjusted based on the cutting model in conjunction with knowledge of the bird's anatomy. The value in accomplishing this work would be to not only reduce labor costs but also to increase the yield of breast meat and reduce/eliminate bone chips. It is estimated that an increase in yield of a single percentage point could represent several millions of dollars of additional revenue for each and every plant. Current attempts at automation of the shoulder cut impose several percentage points of yield loss in return for lower labor costs. In the manual process, while generally providing a higher yield of breast meat, the quality of the product varies dramatically based on the skill of the worker, and the labor costs are significantly higher. It is the goal of this work to develop a system that eliminates labor and consistently provides a yield similar to the best manual worker. The overall vision for this project requires the development of various technology components that will be unified into a single operational system. This includes a system to identify the initial cutting point, a system to specify the nominal cutting trajectory based on the size of that specific bird, a model to predict the location of the joint and shoulder tendons given the position/orientation of the wing tip, a mathematical model of the cutting process that allows the control system to interpret force/torque data and make intelligent motion commands to avoid cutting through the bone, and a robotic platform capable of executing these commands in real-time.
(Georgia Institute of Technology, 2004-02)
Piepmeier, Jenelle Armstrong; McMurray, Gary V.; Lipkin, Harvey
A dynamic quasi-Newton method for uncalibrated,
vision-guided robotic tracking control with fixed imaging is developed
and demonstrated. This method does not require calibrated
kinematic and camera models. Robotic control is achieved at each
step through minimizing a nonlinear objective function by taking
quasi-Newton steps and estimating the composite Jacobian at
each step. The Jacobian is estimated using a dynamic recursive
least squares algorithm. Experimental results demonstrate the
validity of this approach.