The Surprising One Thing that Robots Can't Do as Well as Humans
Apparently, strawberry-picking robots struggle to determine whether a strawberry is ripe or not.
The Surprising One Thing that Robots Can't Do as Well as HumansBy Adam Jezard
Robots have explored the ocean depths, been sent out among the stars and are capable of operating in extreme environments where no human can survive. They can also lift heavy loads and perform mind-numbing, repetitive tasks continually without making mistakes.
But there are some types of human work that robots have not excelled at so far: picking fresh fruit and vegetables.
Fruit picking has posed problems for robot developers, as fruit can be difficult for machines to find and is easy to squash. Robots, unlike humans, have no sense of touch to tell them if they are squeezing too hard. This is a design challenge that is yet to be overcome.
The March of the Autonomous Fruit Picker
Although fruit-picking robots are not as good as humans at this most basic of tasks, steady progress towards automating strawberry harvesting is being made. One US-based company, Harvest CROO Robotics, is developing a machine that should eventually be able to pick a whole strawberry plant in just eight seconds, move to the next in 1.5 seconds and harvest 30 acres in one day.
The company said one machine will be capable of replacing up to 30 human pickers.
In a recent test, a Harvest CROO Robotics machine navigated its way around a field using GPS and a map in its “brain” that shows the location of each fruit-bearing plant, NPR reports. Using sensors the machine recognizes strawberries. The harvester is equipped with high-definition cameras that locate the fruits and robotic claws that gather them.
Picking Up the Pace
“Nobody’s telling [the harvester] what to do,” Paul Bissett, Harvest CROO’s chief operating officer, said during the test. “It’s remembering its path down the row. It's remembering where all these plants are.”
The robots, however, are not as efficient as humans, Bob Pitzer, one of Harvest CROO’s co-founders, said. The robots currently are only able to find and pick over 50% of ripe berries. A typical crew of human labourers can achieve 60-90%.
Despite this, Pitzer is confident that the machines will be widely used within the next few years. The company said its goal is to develop “robots-as-a-service” and make profits from renting out its equipment. The project has already attracted significant capital from the strawberry industry, with up to two-thirds of businesses in the sector investing in the company since its 2013 launch.
Other companies are also looking to take advantage of this growing market, including a company called Agrobot, which is competing to be the first to get its technology fully tested and working.
Rise of the Robot Workers?
While the demand for strawberries and other fruits, including blueberries and avocados, is rising, the number of people willing to pick them – and take labouring jobs in agriculture more broadly – is declining.
Cindy Van Rijswick, sector analyst at Rabobank, in February told website FreshPlaza: “Fruit pickers are hard to come by, especially when it comes to temporary work.
“This is not only true for the US and Europe, but also for a number of regions in, for example, South America.”
Reasons for this include the backlash against migrant workers on both sides of the Atlantic and disadvantageous currency exchange rates, which have encouraged many fruit pickers to return home. Meanwhile, young people have been deterred from working in the fields by the prevailing reputation that fruit picking wages are low.
Some commentators warn that robots will need to harvest a variety of crops for automation to be successful and sustainable. Nolan Paul, head of research and development at California-based berry grower Driscoll’s, said: “I’m still skeptical that harvesting robots are going to be successful if you just focus on one crop.”
Original content can be found at the website of World Economic Forum.
This article is reproduced under the permission of World Economic Forum (WEF) and terms of Creative Commons Attribution-NonCommercial-NoDe