2022 – Will fruit-picking robots change agriculture? | Artificial Intelligence (AI)

sRobots can do a lot. They build cars in factories. They sort merchandise into Amazon warehouses. Robot dogs, a little scary, are supposed to make us safer by patrolling our streets. But there are some things that robots still can’t do – things that seem pretty easy by comparison. Like picking an apple from a tree.

“It’s simple” for humans, says robotics researcher Joe Davidson. “You and me, we can close our eyes, and reach for the tree. We can grope and touch it and say, ‘Hey, this is an apple and the stem is here.’ Pull, turn around. We can do all this without even looking.”

Developing a robotic device that can simply pick up an apple and throw it in the trash without damaging it is a multimillion-dollar effort that took decades. Teams around the world have tried different approaches. Some have developed vacuum systems to vacuum fruit from trees. Davidson and his colleagues turned to the human hand for inspiration. They began their endeavor by observing professional fruit pickers and now work to replicate their dexterous movements with robotic fingers.

Their work can help transform farming, turning fruit picking – a laborious and time-consuming human task – into a quick and easy task for farm workers.

The effort has recently gained momentum as researchers note deteriorating conditions for farm workers amid the climate crisis, including extreme heat and smoke from wildfires, as well as a shortage of workers amid the pandemic. Technology can improve working conditions and occupational safety. But that outcome depends on how the robots are deployed in the fields, farm workers’ organizations say.

While robotic tools for farming have made great strides in recent years, these AI-based tools are primarily used for weeding, monitoring soil moisture and other field conditions, or growing soybeans with remote-controlled tractors. “But when it comes to actual physical labor like pruning trees or picking fruit, that’s still human domain today,” Davidson says.

Teaching robots to perform these tasks requires modern versions of orchards and apples.

Traditional orchards with irregular trees and huge canopies are very difficult for algorithms to analyze and manipulate. Changing sunlight, fog, and clouds adds to the challenges of computer vision. Old, tangled, tall trees are a problem even for human pickers, who end up spending much of their time pulling and laying ladders rather than picking fruit.

Many farmers have now moved to orchards, where trees grow flat on trellises, trunks and branches at right angles to create a “wall of fruit,” says Scott Jackie, owner of Red Roof Consulting, a group that improves cropping techniques contributes. A thin canopy also allows more sunlight in and encourages fruit formation.

Since the 1990s, breeders have been working to develop apple varieties that are more resistant to sunburn — a side effect of those littered leaves — and less prone to bruising when placed in trash cans. All these changes to the trees and the apple itself make things easier for robots (and humans).

In orchards with small trees, human fruit pickers can roam through rows of trees in pairs on slowly rolling platforms. One bends to reach the drooping fruit, the other grabs higher branches. Professionals working in this way take about two seconds to pick an apple.

Essentially a giant arm mounted on a rotating platform, the robot in Davidson’s lab takes about five seconds to complete its movements. With the push of a button, the robotic arm grabs the fruit with its three-fingered palm — actually a plastic apple made for testing purposes. His fingers are covered in a padded silicone skin that hides individual motors connected to the tendons that move his fingers. Thirty sensors under each fingertip track pressure, speed, angle and other aspects of its grip to help the robot do its job.

Another press of the button the fingers clench, then rotate, and the apple – successfully picked – rests in the palm of the robot’s hand.

The fruit-picking robot has succeeded in picking an apple about half the 500 times it has done so far. However, the robotic arm has solved some of the problems that posed obstacles to automation. For example, damage to fruits and branches during the harvesting process can be avoided. Rapid improvements in computing technology have made Davidson and others hope that robots will operate on farms in the next five to 10 years.

The US government is betting heavily on this technology. Last year alone, federal funding agencies awarded $20 million to support the AgAID Institute, a new group that supports several researchers, including Davidson, in efforts to develop AI-based tools for agriculture.

Harvest automation advocates say there will still be jobs for humans, such as training and operating robots. “There are going to be many tasks that will definitely require robotic devices and digital devices to work with humans,” said Ananth Kalyanaraman, a professor at Washington State University and director of the AgAID Institute. “It will actually empower people because it gives them new skills.”

Many farm workers are still not clear how robots will affect their livelihoods. “If used properly, it can actually be a support system for workers and improve working conditions,” says Rina Lopez, executive director of PCUN, a Latin American rural workers organization in Oregon.

But so far, Lopez and others say they have not been involved in conversations about fruit-picking robots. In the past, they say, “farm workers have not been the focus of any of these conversations.” In various industries, including agriculture, waves of automation have led to job losses and the devaluation of human labour. “A lot of times what happens to low-wage workers is that people lose their jobs after these shifts,” Lopez says.

The emergence of robotic farm workers may be an opportunity for humans to pursue other jobs — much less arduous than pruning or harvesting, says Ines Hanrahan, executive director of the Washington Fruit Research Committee. “There are a lot of people in rural communities who can’t do this work physically, even if they wanted to,” she says.

“Excluding the physical aspect makes these jobs more accessible to older workers or those who are less physically able to pull stairs and other things. This allows more people to participate in that work.”