From Unmanned Systems Magazine: Robots and Drones: Farm Workers of the Future?
The agricultural industry is experiencing labor shortages in various parts of the United States today, and some growers and producers are turning to robots and drones to help keep their farms and greenhouses viable.
Immigration policies advocated by President Donald Trump are exacerbating a labor shortage in many regions of the country, particularly California, which is also driven in part by the strengthening of the Mexican economy. The Bank of Mexico recently adjusted upward its forecasted GDP growth in Mexico for 2017 from 1.5-2.5 percent to 2-2.5 percent. In the U.S., farmers in many areas are already dealing with a labor shortage and have expressed concern that migrant workers from Mexico and Central and South America are less willing and more fearful of coming north of the border.
In 2014, a third of the nearly two million agriculture workers were born outside the U.S., according to the Pew Research Center. Immigrants make up about 33 percent of the agriculture industry workforce, with 18 percent undocumented. Immigrants also account for about 60 percent of graders and sorters of agricultural products, with 28 percent undocumented. When compared to older federal data, the Pew numbers show a decline in foreign and undocumented workers in the U.S. industry.
The U.S. Department of Labor’s National Agricultural Workers Survey estimated that between 2007 and 2009, 71 percent of crop workers were foreign-born (67 percent in Mexico and 4 percent elsewhere). The agency found that 48 percent of crop workers surveyed indicated they were not legally authorized to work in the U.S. — a slight reduction from the peak of 54 percent in 1999- 2001.
In California, farm employers are facing major labor challenges, “including new heat safety regulations and requirements to pay piece-rate workers for breaks at their average piece-rate earnings,” according to a 2016 report by UC Davis agricultural economist Philip Martin. “The major labor challenge is fewer Mexican newcomers arriving as unauthorized workers.”
The industry is responding, in part, by turning to tech startups.
California’s $47 billion industry is a breeding ground for field working robotics. Tech startup Agrobot, based out of Spain, is working with the financial backing of berry producer Driscoll’s to make its strawberry-picking robot commercially available next year.
“The problem right now in California is that they don’t have the labor,” Agrobot CEO and founder Juan Bravo tells Unmanned Systems. “They have to close the farms or move the farms to Mexico. The problem is not to replace the labor, the problem is the labor is not there.”
Bravo says Agrobot can reduce labor costs by about 15 percent, according to the company’s current estimates. Bravo says he could not provide more detailed pricing.
The Agrobot uses computing software, as well as sensors and cameras, to identify a strawberry and, using a morphological and color analysis, register the ripeness of the fruit to meet the standards of the farmer. This system analyzes the fruit one by one, “and it is responsible for ordering cutting movements that guarantee accuracy, smoothness, and sensitivity in the strawberry treatment.”
Sixteen robotic arms cut and take the fruit, placing it on a conveyor where a human worker places it in its package.
The technology is not without its challenges. With strawberry fields being an outdoor crop, human workers are more agile and flexible in incremental weather conditions than robotic machinery. Dust and rain can cause problems, but addressing those problems can increase the cost. Instead, Bravo said the machines need to be cleaned weekly. The strawberry harvester also relies on electricity, which must be accommodated with wiring, a battery or an electric generator.
“The simplest solution is to use a generator but some clients for indoor applications will want to use batteries,” Bravo says.
Spain-based Agrobot is working with berry giant Driscoll's to market its strawberry-picking robot. Photo: Agrobot
Non-food crop industries
Harvest Automation, based out of New Hampshire, has similarly seen a major push from growers and farmers to accelerate automation of the agriculture industry. Harvest, however, learned early on of the challenges of automating outdoor crop labor, and has targeted non-food crop industries.
“That's one of the significant reasons that we chose this nursery greenhouse application is they have limitations on the terrain conditions for their own purposes, like for growing plants, it can't be too rough,” Harvest CEO and cofounder Charlie Grinnell says. “It can't be big mud puddles and standing water everywhere because that promotes disease in the plants.”
Harvest has been producing and selling the HV-100 at a cost of about $3,200 per unit since late 2013. Using sensors and behavior-based robotics, the HV-100 is in charge of spacing and aligning potted plants when they are potted, when they begin to grow, when they need to be protected from incremental weather, when the season changes and when they are ready for shipment. The HV-100 also collects data to inform grower operations.
Grinnell said each HV-100 replaces about one worker. But when you account for the costs of management supervision, employee benefits, workers compensation and the robot’s ability to work longer hours with fewer mistakes, the robot pays for itself in about two years, depending on the job, he says.
With financial backing from a number of growers, the tech startup saw early on the industry’s demand for automation to supplement a decline in available workers.
“Some of our customers don't have really have any labor problems at all, there's plenty of immigrant labor or part- time labor,” Grinnell says. “But other parts of the country there's just no labor. Even with a move like California has done to increase the minimum wage to $15/hour, some of our customers still have a huge challenge in just labor availability.”
California last year passed a law to raise the minimum wage to $15 an hour and while this has helped attract more people to jobs in the agricultural industry, the shortage of workers persists.
But Grinnell also sees industry challenges in getting the venture capital firms to invest and put his company through development and commercialization. Because the agricultural industry is broken into several different types of crops, there’s no one-size-fits-all robot for strawberries, shrubs, grapes and asparagus. Each requires its own unique robot, if the industry is going to replace humans with machines.
“You just can't demonstrate that you're going to have a big market for this product in the end of the day,” Grinnell says. “That was a big challenge, but I was able to convince enough of them to part with their money that we were able to get the product developed.”
Grinell says it helps that greenhouse and nursery growers operate year-round, while many other agricultural crops are seasonal.
Harvest Automation is, however, in talks with the poultry and grape industries to develop material handling robotics using the company’s existing platform.
Blue River Technology, based in California, got its funding by nabbing a big fish. Deere & Company, maker of John Deere agricultural products, on Sept. 6 signed a $305 million definitive agreement to acquire Blue River, which already provides drone technology to inform plant breeders and a robot to thin lettuce in the fields. Blue River also received $17 million in funding from Monsanto and Syngenta about two years ago.
Blue River is less focused on replacing labor than it is on optimizing data and efficiency. For example, the company markets its drone imaging system to plant breeders to collect data over field plots using sensors and a multispectral camera system to collect and analyze data, Blue River Senior Systems Engineer Matt Colgan says. That kind of work is usually delegated for 10 days at a time to groups of high school students prone to data entry errors. The use of sensors and cameras also provides data at a level that the breeders did not have before. Blue River employs commercially available unmanned aerial systems and equips it with its own patented technology for the field readings.
The company’s ground-operated LettuceBot, however, may be replacing actual labor. LettuceBot thins lettuce while it’s growing to keep the plants healthy.
“The cost of the thinning services is comparable to the grower's cost for labor,” Blue River says. “Operation of the robots may offer a direct cost reduction, a management cost reduction, and also a yield increase.”
LettuceBot uses machine learning to identify the plants and precisely trim each head of lettuce. Like the HV-100, it is also trained to avoid obstacles.
Blue River is also working on making commercially available next year its See & Spray computer vision and machine learning technology to discern weeds from cotton and target and spray them with herbicides with a level of accuracy that leaves the cotton plant untainted. Colgan says the technology, by directly targeting the weed, reduces the use of pesticides by 90 percent compared with aerial pesticide spraying, usually conducted by an airplane at a cost of $1,000 an hour. Deere’s investment in Blue River supports its efforts to employ similar technology on a wider range of products.
“They really want to begin to incorporate computer vision into all of their machines,” Colgan says. It not only allows for precise automation but data readings can improve the accuracy of John Deere’s existing harvesters.
The agriculture industry is a long way from full automation, and some of the smaller crops may have a tough time financing it, but major growers and breeders are realizing benefits from automation not only in reduced labor costs over the long-term, but potentially much higher yields.
Below: Harvest Automation builds robots that can sort potted plants. Photo: Harvest Automation