Farmers need robots. They are called digital farmhands. Persistent, resilient machines that can quickly pick fruits and vegetables are one of the few ways to exploit increasingly small windows of opportunity between weather events.
But it takes a village to keep such complex machinery running.
Professor Salah Sukkarieh, a robotics engineer at the University of Sydney’s Australian Centre for Field Robotics, says agricultural robots won’t be replacing farm labourer’s jobs any time soon, but they will become a necessary tool to harvest as much as possible, as quickly as possible.
Earlier this year, Sukkarieh addressed the United Nations Global Conference on Sustainable Plant Production on the ability of agricultural robots and AI to boost productivity and yields, and improve food security.
But he’s told Cosmos that they will also influence the human economic ecosystem around them.
Farmers are already jacks of all trades, but adding robotics to their CVs might be too much to expect.
This means regional towns and cities will need to adapt.
Just as blacksmiths evolved into motor mechanics, another technological era is unfolding.
But will local businesses fix, adapt and maintain these machines, or will they be bypassed, with the faulty digital farmhands being sent overseas for servicing?
Professor Sukkarieh says there’s currently a struggle between two philosophies.
One calls for autonomous farming machines to be as open-source as possible. This means they will be designed from the ground up to use commonly available components, sensors, computer processors and software.
“So the open source camp basically says that people like your local mechanic and the local software guy should be able to deal with the systems,” says Sukkarieh.
“You must design the robot from the ground up to make this work. You’d be using whatever you can find in regional areas, anything from standard motors, chains, sensors – anything you can buy off the shelf.”
The other model is a closed intellectual property loop like smartphones and tablets.
“This camp believes that the only way these machines will be economically viable is to retain control over the technology and software, to interact with the farmer directly by leasing the equipment and offering a service contract.”
Sukkarieh says he expects the two movements will likely converge.
And that’s because of the broad variety of technology and artificial intelligence needed to tackle farming roles.
All robots have basic drive components, such as electric engines, power packs and wheels. These will always need mechanics to service them.
Professor Sukkarieh’s University of Sydney Institute of Agriculture team has developed a low-cost, autonomous multipurpose robot to assist on smaller farms. Called Digital Farmhand, it’s essentially an intelligent electric tractor.
“Digital Farmhand is a small, autonomous electric tractor-like vehicle that can tow a variety of implements such as seeders, weeders and bed preparation tools, and can undertake precision automation of many labour-intensive farm tasks, like weeding, spraying and seeding,” he says.
“Digital Farmhand can also use accessible smartphone technologies along with AI to provide crop analytics such as yield estimation or pest and disease identification.”
An example of an open-source platform Digital Farmhand is one being envisioned for small-holder farmers across the Asia-Pacific. This uses electric and petrol scooter parts commonly found in the region to make spares easier and cheaper to find. And open-source applications on smartphones can access its artificial intelligence.
More in Cosmos: Academy lets anyone build robots
“But once you get to the part where you start talking about computing and sensing – the functions that turn these little tractors into robots – that’s where things will probably have to be taken out of the community’s hands,” he adds.
Grape-picking robots are much more complex. These require greater computing power to process finer-resolution vision algorithms to guide sensitive robotic arms and hands.
“We don’t build robotic arms in Australia, “Sukkarieh says. “But we do build mechanical weeding tools for robotic platforms. So it’s all about how complex any given part is.”
Currently, most agricultural robots are active in the fields and orchards of “early adopters”.
Farmers are like any other gadget consumer, Sakkarieh says. Some are eager to experiment with new ideas and tolerate the challenges of getting them working properly. Others wait for a more polished product to come online.
Ultimately though, it’s a numbers game.
“What they’re looking for is how much fruit you’ve left on the tree, how many pieces you’ve damaged, and how quickly you collected the rest,” Sukkarieh explains.
A robot doesn’t have to be perfect.
“Let’s say it picked 80 per cent of the fruit and damaged 1 per cent. That leaves just 19 per cent of a crop needing human labour to harvest. And that could be a profitable trade-off for a farmer.”
But raw economics remains a challenge.
A grape-picking robot may be frantically busy bringing in a harvest for a month. But what does it do for the remaining 11?
“Farmers can’t afford idle assets,” says Sukkarieh. “So we may end up with travelling robot picking companies following ripening crops.”
That means some agricultural robots will have to be adaptable, with software and gripping hands capable of picking oranges one week and avocados the next.
Others will find more steady employment.
Strawberries, for example, are harvested year-round. And their farms are very uniform and regulated environments.
“The more structured the environment, the better a robot’s performance,” Sukkarieh says.
Are you interested in how science and technology is transforming production, energy, and agriculture? Then our new email newsletter Greenlight Project, launching soon, is for you. Click here to become an inaugural subscriber.
The Greenlight Project is a year-long look at how regional Australia is preparing for and adapting to climate change.