Artificial intelligence is being touted as an answer to falling productivity in many industries, and agriculture is no exception.

Many applications of this groundbreaking technology might tackle complex problems around yields, managing inputs and even
tackling emissions, but there is room for AI to help in the more mundane but important parts of farming.

Global farming technology company GEA collaborated with local start-up AgriAI to launch a new system called the walkover teat sprayer at Fieldays last week to combat the problem of mastitis in dairy herds.

“Mastitis costs our farmers a lot of money,” GEA Farm Technologies NZ product manager Ben Morris told the Herald at Mystery Creek.

“Cow’s milk actually goes down the drains. We want to improve outcomes for our farmers.”

The bacterial infection creates inflammation in the udder, reducing milk yield, and is contagious. For decades farmers have tackled the problem by manually spraying at milking time.

Despite efforts to contain the disease, it’s estimated mastitis translates to a loss of about $180 a cow a year.

The problem is costing the New Zealand dairy industry hundreds of millions of dollars a year.

Benedict Johnson and Chris Scherman, recent graduates from Waikato University, set up AgriAI to target mastitis.

“Cows might have 20 incidents of mastitis in a year,” Johnson said. “If we can improve that, then we can improve the overall profitability of the farm.”

Johnson grew up on his parents’ dairy farm on the Coromandel Peninsula, but was drawn to mechanical engineering and graduated from Waikato majoring in mechatronics and robotics.

At university he met Scherman who was studying electronic engineering, and the pair discovered a mutual interest in AI.

“We share an interest in machine learning and think it’s the way of the future,” Johnson said.

“We had an understanding of what else was out there on the market and going back to first principles we thought — how do we actually automate this procedure?

“The origin of the teat spray was an understanding of the limitations of other teat sprayers in the market, especially around coverage, usage and effectiveness of the teat spray.”

And that accuracy of application is affected by animal behaviour and accidental interference.

“Cows behave in some quite odd ways,” Morris said. “Sometimes they push each other through the race, sometimes they run quite fast. Because we’re monitoring in real time, we can really accurately apply the teat spray.”

The system uses a high-mounted camera that observes cattle movement in real time and uses that information to decide when to apply the spray.

“The camera above is to keep it clear of weather and other potential obstructions,” said Johnson.

“This is a crucial part of machine vision to keep the lens clear, so that you can detect the cow.

“We have a neural processing unit that takes the images from the camera and identifies whether it’s a cow or another thing, and then it will track the cow and then very accurately apply the spray.

“As the animal walks through, once the udder comes over the spray pad, we use the four nozzles to apply the spray. The camera can detect when there’s any dirt or mud on the nozzles and instruct the unit to self-clean — the system is intelligent enough to be self-cleaning should manure get on to the spray nozzles.”

Johnson said with the various levels of automation on farms, he believes the walkover sprayer easily fits with the simple farming systems in New Zealand.

“It’s a tack-on piece of equipment that fits with our customer base,” he said.

Morris said the device automated the process without compromising the health of the animal — and the farmer got a good productivity gain.

“I think they’re looking for productivity gains on farms,” said Morris. “Labour is becoming harder and harder to find [and] I think we’re solving some problems for them [farmers].”

And, although he’s tight-lipped on what’s next for AgriAI, Johnson believes machine learning and AI will be “massive” for agriculture.

“There’s definitely some machine-learning technologies that we can use this real-time practice for,” he said.

“I’ve been working on this for about two years now and in partnership [with GEA] for about a year. The difficult parts are around making the real-time machine vision cheap and reliable and all the software in behind that. It’s been a huge task to get that together.

“You see the environment with your eyes, so we’re giving machines that same capability and in agriculture it will really benefit farmers.”