The matter of growing enough food to feed the planet is a serious issue. Already, one in nine people lack sufficient sustenance, according to the United Nations—and the problem is only going to get worse. Mix a global population projected to reach 9 billion in 20 years, from 7.5 billion today, with drought and other effects of climate change, and a farm labor shortage in places like California, and there’s a crisis in the making.
Now a growing number of researchers and are turning to robotics to address the problem. Specifically, they’re combining the sensing abilities of robots with data analysis made possible by artificial intelligence technology to improve farmers‘ ability to grow and manage their crops more intelligently. “We can gather information in ways we never could before and use machine learning techniques to analyze these large data sets and help growers make more intelligent decisions,” says George Kantor, senior systems analyst at Carnegie Mellon University’s Robotics Institute.
At the same time, however, there‘s still a long way to go before these systems are sophisticated and capable enough to make a big impact—and farmers are willing to adopt them in a mostly low-margin industry. “It’s early days,” says Kantor. “And farmers tend to be very conservative.” One area where there’s more interest, according to Kantor, is in higher-margin, specialty crops like orchards and vineyards, which are more labor and management-intensive than, say, corn or soybeans.
Research reaching a critical mass
For now, a lot of the action is happening among university researchers like Kantor. He’s involved in FarmView, a series of long-term projects aimed at improving plant breeding and crop management techniques, working with agriculture-oriented universities ranging from Texas A&M to Washington State. While he’s been involved in such work for 20 years, recently, says Kantor, “We’ve started to reach what I would call a critical mass.”
Working with researchers at Clemson University, they’ve recently made great strides with a grain called sorghum, a coarse dry grass with more than 42,000 varieties. Robots travel between rows of sorghum collecting data for further analysis. Specifically, the robots have grippers that can poke needle-like devices into stalks to measure their strength, and then take a picture. The aim is for software to then analyze which ones should be cut or saved.
So far, most of FarmView’s work has focused on using robots and AI technology for data analysis and gathering. The next step is for robots to do something, suggesting action steps or even taking those actions themselves. For example, researchers recently worked with vineyard owners concerned about “line balance”, or how many leaves vs. grapes are on a vine. They’ve developed robotic systems that measure and count every grape and leaf, then create maps denoting areas where there are too many of each item. Ultimately, with an automatic management system, machines can enter the appropriate areas and remove the offending items themselves.
Researchers, as well as companies, are also applying robots and AI to weeding, especially in reducing the amount of herbicide used by farmers and also guarding against overuse, which can lead to resistant strains of weeds. Traditionally, they’ve had to spray an entire field, even though most have weeds covering about 10% to 20% of the area, according to Ben Chostner, vice president of business development at startup Blue River Technology. Farmers have created genetically modified crops able to withstand a limited number of herbicides. But Blue River’s system allows robots to identify all the plants in a field, then, using thousands of images, teach them to recognize which ones are weeds. Thus, famers can plan their activities down to the individual plant level, avoiding wholesale spraying of an entire field. Because they can pick and choose which plants to target, they’re able to use a wider variety of herbicides.
In 2013, the company, formed five years ago, started with a system aimed at lettuce. Now, it’s introducing one for cotton, which has perhaps the biggest weed control problem of any plant. “These guys are out of tools to control their weeds,” says Chostner.
Another hotbed of activity: The agBot challenge in Rockville, Ind. Now in its second year, the competition pits teams of entrepreneurs and researchers applying robotics, AI and other advanced technologies to farming. For example, this year, 10 teams competed in the “Weed & Feed Competition”, for which they developed robotic systems trained to maneuver around rows of crops, determine which ones needed to be fertilized, identify three common weeds, arrange for them to be eliminated either chemically or mechanically and relay real time information about how to fertilize or treat the plants back to a base station. “They’re allowing farmers to make better decisions,” says Rachel Gerrish, co-founder of the challenge.
The contents or opinions in this feature are independent and may not necessarily represent the views of Cisco. They are offered in an effort to encourage continuing conversations on a broad range of innovative technology subjects. We welcome your comments and engagement.
We welcome the re-use, republication, and distribution of “The Network” content. Please credit us with the following information: Used with the permission of http://thenetwork.cisco.com/.