How the AI 'cow-gorithm' is helping this dairy couple get the job done

Moving the cows on this farm to the milking sheds has never been easier.

Simon Badcock can open an app on his phone and set up some virtual fences that guide cows all the way to the dairy shed.

And it doesn't stop there.

Once the cows arrive, robotic milkers automatically attach and begin milking.

The farmer and his wife, Sonia Badcock, can be in a cafe having a coffee while the work gets done.

"You can look after the rest of the farm better, and you can have time off," Mr Badcock said.

The couple's farm on Tasmania's remote north-west coast is the first in Australia to use virtual fencing collars and robotic milking machines together, and has done so for two years.

These days, Mr Badcock doesn't have to muck in quite as much.

He said he just needed to keep an eye on things, check his herd, and wash down the machines, "otherwise we might not be there at all".

The Badcocks only have to be present if a robot or collar flags an immediate issue.

"Because we're pushing it to its limits with nearly 100 cows per robot, we can't have too much time with the robots being down, so we come during the night. Once a week, maybe," he said.

"We're making improvements as to how efficient we can run the farm with those technologies, and it is definitely getting better and easier all the time."

Using the 'cow-gorithm'

The GPS-enabled collars being used track the location and health of each cow on his property and move cattle with their patented AI "cow-gorithm".

The machine learning algorithm uses vibration, sound, and herd behaviour to direct animals between invisible paddock boundaries.

If an animal does not respond, it delivers a low-energy electric pulse.

Ms Badcock said their operation took advantage of this to move batches of cows to their dairy, where the animals' instincts then took over.

"They know that when they go into the robot, they get grain, and then when they leave the robot, they get fresh grass,"

she said.

The robot identifies each cow from its ear tag and decides whether to milk it.

If there hasn't been enough time between milkings, the animal eats and moves on.

"The robots are open and on all the time, so as soon as a cow walks into a robot the gate closes behind it and the robot … knows which cow it is and just starts milking," Ms Badcock said.

Dr Megan Verdon, a researcher at the Tasmanian Institute of Agriculture, has studied the welfare implications of GPS collar systems, and said that since being introduced to Tasmania, virtual fencing had helped reshape labour and infrastructure for the sector.

"It lets us ask, 'if you could remove those barriers, how would we have designed our system'," she said.

Dr Verdon believes that research into the use of virtual fencing with autonomous cow-driven dairies is needed.

"Any system has trade-offs," she said. 

"That system works with the cows by bringing them all in as a mob, but the cow's not choosing when she's milked, which is where these robot systems were developed and [where] people start talking about agency.

"That might be okay, we haven't looked at it."

As cows queue in a yard to go through the robot system, Dr Verdon said it was similar to a conventional system.

"The biggest difference I can see from Simon's system, where he uses it to batch milk cows, to a conventional system, is not having a human in the mix, which could be good or bad," she said.

"You still need to know animal health, but it may reduce the load on the farmers who are already working 12-hour-plus days."

Currently, Halter collars are approved for use in all states but Western Australia, where they are being assessed.