The cause is known as Moravec’s paradox. Machines are wonderful at doing things humans struggle with, like quantum physics, but struggle with basic physical actions that humans find simple.
“It seems like one of the most intuitive things that humans do, you take a person to an orchard and they just pick an apple,” says Leopold Lucas, co-founder of another start-up called Ripe Robotics, which working on a machine to do just that.
For a robot, the challenges are many. Unlike in a warehouse, where automation is becoming commonplace, there are no smooth surfaces for a robot to roll along in an orchard. There is no road map to follow. No two apples look exactly alike, and no two trees grow alike. This means that every time a robot goes to pick an apple, it must be smart enough to analyze the entire tree and understand what it is reaching for and how to get there without getting stuck on a branch. Then Eve, Ripe’s current test robot named after the biblical picker of forbidden fruit, must apply just enough pressure to get the fruit off the tree without leaving even the tiniest mark that could prompt supermarkets to drop the price by 90 percent. And that’s before taking the weather into account.
“You have to do it out in the field, where there could be wind, rain, snow, 40 degree heat. It will be difficult, very dusty,” says Lucas. “There are difficulties in identifying the fruit and assessing how ripe it is, especially under different light conditions.”
After all, a big selling point for robots is that they can work in conditions where humans can’t, like in the dark, and for more hours than humans, who need food, drink, rest and toilet breaks. There can also be advantages in tracking the provenance of food and analyzing its quality, but all this means is that robots must have the durability and the necessary power supply to last a long time.
Conditions are not as complex in a packing shed, but the basic challenge of quickly grabbing and moving unevenly shaped fruit remains.
Lyro co-founder Dr. Nicole Robinson, who has a background in robotics at Queensland University of Technology, says there is a degree of difficulty. Common, roughly spherical fruits like apples are among the easiest for a machine to grasp. Sweet potatoes, which come in a variety of sizes and twists, are in the middle. Small and picky berries are some of the hardest.
But there have been a host of innovations in recent years that make the challenge manageable. Computer chips have become much faster and more efficient, let robots perform more complex tasks. Vision systems that let them identify objects have improved dramatically. Internet in the countryside is faster and more reliable. And even some farming practices—like planting apple trees on trellises that make orchards easier for machinery to navigate—help, too.
Robinson says the combination of advanced technology with a big problem to solve is what drew her to the sector. “There is an enormous amount of food waste that occurs if you cannot pick and pack the fruit in time. You know, millions of tons are just wasted a year.”
Food loss on farms has a number of causes: disease, market prices and weather among them, as well as insufficient labor, but Robinson is right about the cost. A 2015 report by a government industry-academic research center estimated it at almost $3 billion a year in Australia alone.
There is also, says Lyro’s general manager Mark Adams, the opportunity to move workers from the hard, boring and repetitive tasks on farms to higher skilled and better paying jobs. This can mean things like analyzing products, overseeing machinery and performing skilled maintenance. Boardman hopes it will go that way as well.
There are plenty of roles in the former category.
It’s someone’s job to put a blob of hot glue on each of the thousands of pineapples that come down a conveyor belt for eight or 10 hours a day, and someone else’s job to put a sticker on it, Adams says. It is someone’s job to put e.g. 42 peaches in each tray: no more, no less. Someone must cut each cabbage, bent forward in the heat or the cold. Laborers and machine operators, who are common on farms, have some of the highest rates of serious work-related injuries of any profession, Safe Work Australia statistics show.
And on unscrupulous farms, underpayment is widespread. In 2018, audits by the Fair Work Ombudsman found that more than 200 firms in the industry broke pay laws; Oh that checked again in 2020only one in five had solved their problems.
But there are also advantages to agricultural work. It helps sustain regional centers that, at least before the pandemic, were struggling with population loss. And governments from both sides of politics have championed the deals Australia has with Pacific island nations to grant visas to farm workers, arguing they give workers skills, cash to send home to family and are a way to deepen ties in the region at a time when the competition for influence with China is fierce.
The United Workers Union, which represents horticultural workers, declined to make an official available for interview.
Even if Adams is right, and farm workers are becoming more skilled, it seems that automation will mean fewer in the future, at least of the seasonal workers. The economy does not work in any way.
So far, Lyro and Ripe are small firms, with the former’s last raise totaling about $1.5 million in capital and grants, while the latter has achieved about $1.2 million in larger raises and equity.
Both plan to raise more money from investors to keep developing their robots, despite the difficult conditions for raising seed capital, with Lyro estimating that it will need a few million more to get to a larger commercial scale to rent out to farmers at a planned cost of about $7000 to $8000 per month. That model means they can move them around from farm to farm as different crops come into season.
Ripe is experimenting this year in agreements with farmers, where it is trying to match human costs of about $60 per container of picked apples. Once it proves its concept by picking a full bin, Lucas says Ripe hopes to raise $3 million to $5 million to hire staff and expand its fleet of robots.
Boardman, the avocado farmer, says he will only use robots on a large scale when they are as fast as a human and comparable in cost.
“I think we’re going to find that it’s going to take some time to get robots up to human speed,” Boardman says. “What the human body is capable of doing is truly amazing compared to what robots can still do.”
The morninging Edition newsletter is our guide to today’s most important and interesting stories, analyzes and insights. sign up here.