Feb 18, 2020

Technology is on the move for growers

It sounds like science fiction: Robotic vehicles moving up and down orchard and vineyard rows, gathering information to guide growers in everything from harvesting to disease detection and pruning.

Yet, that technology is under development right now, according to George Kantor, senior systems scientist with Carnegie Mellon University in Pittsburgh, Pennsylvania. Addressing the Washington State Tree Fruit Association’s Annual Meeting & Horticultural Expo in December in Wenatchee, Washington, Kantor outlined how this cutting-edge technology will help growers.

For a number of years, automated vehicles have existed, sometimes rigged with cameras and large lighting equipment to take pictures of fruit. A camera card would then be taken back to the research lab for processing. However, rapid advances in computer technology now make it possible to produce sophisticated 3D models, with information directly available from the field, said Kantor who works in the Robotics Institute, a division of the university’s School of Computer Science.

“There’s a special kind of computer system – GPU (Graphics Processing Unit) – that makes computer games work so well,” he said. It’s also really good for work in the field because it is small, lightweight and is becoming less expensive, he said.

Today, “if you can train a computer to do that,” Kantor said.

In the orchard or vineyard, the three key words are: mobility, perception and manipulation, he said.

In past years, “mobility” might mean just putting a camera on the back of an ATV oreven putting it on a wagon “with grad students pushing it around,” Kantor said. Today, “autonomous mobility capabilities” have made a big difference. Now, researchers use robotic vehicles, that can be programmed to move up and down vineyard or orchard rows, even navigating bumps and ditches.

“Perception” of the system has also markedly improved.

A more efficient, less cumbersome unit has been designed to mount on the robotic vehicle, containing a camera, lights and computer.

In a huge grape vineyard, for example, the network can be trained to count grapes and determine the size and color of fruit along the rows.

Measurement, predictability

“You can take measurements in July and predict the yield for the harvest that’s going to take place in September,” he said. By assessing how ripe the grapes are in certain parts of the vineyard, it’s even possible to determine where to send workers first, to begin harvest. The technology also can measure the canopy in an orchard and determine leaf area size. Today, these more advanced systems take good pictures in any lighting.

In an orchard, robotic vehicles can even scout for disease, finding spots on leaves, for example, Kantor added, in a conversation with the Fruit Growers News.

The “really challenging” part of this new technology is the element of “manipulation” – or “creating intelligent mechanisms” for harvesting and other labor-intensive tasks to help growers, Kantor said.

Carnegie Mellon researchers are continuing to make strides in this arena, with ongoing innovations. For example, by developing a smaller version of a camera which can be held in a robot’s “hand” it will be possible to get up closer to trees and vines for making pruning decisions.

Studies are currently underway to use data received in bud counting to help a robotic system prune in the most efficient manner. However, the fine-tuning necessary, such as having a tree shaped in a certain way to accommodate robotic machines, is tricky.

“It’s going take a long time,” Kantor said, to have the complete robotic system readily available to growers. “This is coming to your industry, not next year, or even in five years, but it’s coming,” he predicted.

Carnegie Mellon is not doing research with drones, although sometimes researchers coordinate with those who use drones to incorporate images they capture into an overall data bank, Kantor added.

A disadvantage to drones is that they “will always be above (the orchard canopy),” he observed. “You can’t see up under the canopy.”

Kantor encouraged growers to begin to lay groundwork for emerging robotic technology. Keep fruit as visible as possible, perhaps by transitioning to a 2D fruiting wall if you haven’t already done so. Keep your terrain relatively flat to accommodate the robotic equipment. And keep your growing systems “really regular,” with trees and fruit evenly spaced in systematically straight rows, he recommended.

“The closer the orchard looks like a factory, the better the robots are going to do,” he observed.

Grower buy-in essential

This innovative technology still needs “millions of dollars” to move forward, Kantor said. If growers show interest, it will demonstrate to potential investors that “there is a compelling idea and that there is support of customers.” When a grower buys equipment or participates in a pilot program, “it is a vote from you,” he said.

Check with local university Extension agents for projects to participate in, Kantor suggested. There are also a number of startup companies beginning to provide products and services in this area. There is already robotic equipment available “at an early commercial stage” that may be mounted on an ATV and driven through fields to do fruit counts, he said. Marketing also has begun for “mobility units,” to make it possible to program vehicles to carry fruit through the fields.

“I want you to think about more than just the bottom line,” he said. “What you’re doing is investing in a capability which might not make it to viability without you or without customer support.

“The long-term place we’re going to is to help (growers) with their labor concerns … these are intermediate building blocks to robotic harvesting,” Kantor said. “It’s a long path to robotic harvesting. The way we get there is by doing all of these incremental things along the way.”

Ideally, one day these developing technologies will help reduce growers’ uncertainties about an available labor pool, provide up-to-date information on what’s currently happening in the field, and help to guide the “right management decisions at the right time,” he said.

— Christine Corbett Conklin, FGN correspondent; top photo: Robotic harvest-assist units, along with crop monitoring systems being deployed with drones are among the technologies being pursued in the specialty crops industry.