These robots are designed to work with existing rotary parlors
BY Wilfried Wesselink
The ‘holy grail’ of robotic milking systems are machines that work with existing milking systems with a minimum of retrofitting and added expense.
New Zealand engineering firm, Scott Milktech, is well on its way to reaching that dream. The robot it is developing works with an existing external rotary parlor and is used to attach teat cups only. Automatic take-offs remove the units once milking is complete.
The robot is positioned next to an ordinary outside Waikato rotary with automatic take-offs. The platform rotates during the whole process.
In the rest position, the milk cluster is hanging upside-down on the cord of the automatic take-off. When a cow rotates into position, the arm of the robot moves to the milk cluster. The "hand" of the robot, with four pairs of fingers, simultaneously grips the four individual stainless steel teat cups.
At the deck rotates, the robot arm moves the milk cluster to the left and turns it upward. The arm then stretches out to move the cluster between the hind legs and under the udder. The camera on the tool simultaneously detects all four teats, and each cup is individually lifted and attached in order: front left, front right, rear right, rear left.
Placement on narrower teats seems to be no problem because the tool adjusts to the right or left. While doing its job, the robot arm is moving with the rotary. Once all four teat cups are attached, the robot moves to the rest position of the cluster at the next stanchion and starts the same process again.
"The camera on the robot starts by looking at a large generic area. It is measuring a quarter square meter at the back of the cow, then it tries to recognize the udder, and finally it tries to recognize the individual teats," explains Chris Hopkins, CEO of Scott Technology Ltd.
(ScottMilktech Ltd. is a partnership between Scott Technology, which is a 61% joint venture owner, and Milktech Ltd., which owns the other 39%.)
The robotic system only attaches teat cups; it does not (pre)check milk for quality or other factors. That should be done separately by a person or electronically by the milking system. Hopkins says a system could be added that cleans the udder before the cupping robot begins, or dips or sprays the teats after the milk cluster has been removed.
A tail deflector is installed for the whole parlor to make sure the tail of the cow is not hanging in front of the camera or on the teat cups. The tail deflector is a metal bar bent like a double hair pin—it catches the tail during teat cup attachment and moves it aside.
The robot attaches the milker cluster to the cow as the rotary platform turns. After each attachment, it returns to the start position for the next cow.
To attach the four teat cups, the robot currently needs 12 to 22 seconds, averaging 15 seconds per cow, Hopkins says. A human milker in the same rotary needs six to eight seconds to connect all four cups.
The eventual goal is for the robot to attach all four teat cups in 10 to 12 seconds. The robotic attachment time is mainly dependent on whether a cow is moving since the robot moves along with her. If in 22 seconds all four teat cups are not connected, the robot stops trying, and the automatic take-off puts the milk cluster back in the rest position.
Then there are two options: Either the cow will move around on the platform for a second attempt, or a person must attach the teat cups. The cupping success rate on first attempt is currently 95%, Hopkins says.
The robot interface can stop the rotary in case of emergency. The system can also be adjusted to let the platform rotate more slowly or stop it when cupping a cow is problematic. The robot can also get information from electronically recognized cows, for example, such as approximate teat coordinates. "When the robot knows the coordinates of a cow’s teats in advance, it makes on average only half-a-second-per-cow difference to the speed of connecting the milk cluster," Hopkins says.
The Scott Milktech system can be installed in new and existing rotary dairy parlors of various brands. Requirements are automated take-offs, an automated drafting system to separate cows, access to compressed air, power and Internet connection for diagnostic purposes.
The system is designed mainly for external rotaries because there are only a few internal rotaries in New Zealand. "As long as teat cups can be attached through the hind legs, the system would also be available for inside rotaries," he adds.
The system is designed for rotaries with 50 to 120 stalls. With 120 stalls, two robots and one person would be needed. In that situation, the maximum capacity for each robot will be 240 to 300 cows an hour when average connecting time takes 15 seconds per cow.
The optimal rotary size depends on the speed and cost of a system. But Hopkins believes a rotary with 50 to 80 stalls, with one robot for 500 to 600 cows, would work.
"Our robot is meant to assist the farmer, not to replace him," he says. One person would still be required to monitor the system.
Hopkins expects the system will be commercially available by late 2014 or early 2015, first in New Zealand. "The price of the system will depend on the size and required modifications of the parlor and will probably be somewhere between $200,000 and $400,000," he says.
In New Zealand, Hopkins expects that four to five years after investment, the annual cost of a robotic system will be cheaper than a human milker.