No matter how green a farmer's thumb, keeping weeds at bay is a Sisyphean undertaking—one that usually involves manual labor, toxic pesticides or a combination of the two. Weeds compete with crops for space, water, light and nutrients and, if they're not adequately controlled, reduce yields and lower profits. But the standard means agricultural producers have to control them—herbicide application, mechanical and manual weeding—are costly, too.
Take a high maintenance crop like lettuce, California's largest vegetable crop, for example. A report from the UC-Davis department of agricultural and resource economics estimates that weeding and thinning the plants comprises 27 percent of the total per acre growing expenses—much of that reflecting the cost of manual labor.
Enter Blue River Technology, inventor of the Lettuce Bot. The company's ultimate goal is to automate the process of distinguishing crops from weeds, targeting the latter without damaging the former and without using chemicals. "The problem with herbicides is that not only do some other countries refuse to purchase food from the U.S. grown with pesticides, but the whole trend toward organic and natural food and farming is moving farmers away from using toxins," says Blue River CEO Jorge Heraud, MS '96, MSM '11.
The prototype system uses cameras, computer vision and machine-learning algorithms developed by Heraud's cofounder Lee Redden, MS '11. Pulled behind a tractor, the device scans passing plants and compares what it sees to a database of more than 1 million images. When a plant is identified as a weed—or as a lettuce head growing too close its neighbor—a nozzle delivers a spritz of fertilizer, which, at a concentrated dose and close range, turns out to be a highly effective herbicide. The battery-powered system crunches the image data fast enough to work with 98 percent accuracy while chugging along at 1.25 miles per hour.
Blue River has raised $3.1 million from Khosla Ventures (founded by Vinod Khosla, MBA '80), and is field-testing the Lettuce Bot in the Bay Area. But lettuce is just the tip of the iceberg, so to speak. Heraud and Redden have their sights set on the nation's biggest crop: corn.
Meanwhile, another Stanford-incubated start-up called Solum has developed a smart soil analysis system that can help reduce waste by telling farmers exactly when and where to apply fertilizer. Most crops need supplemental nitrogen—a key component in common fertilizers—at some point in their growth cycle. Yet factors including soil texture, field slope and drainage can affect how much nitrogen reaches crops and how much is released into the air, where it acts as a greenhouse gas.
"While organic farmers were using different forms of nitrogen than traditional farmers, there was still a huge issue around how much was safe to use," says Solum president Mike Preiner, MS '07, PhD '09. When the National Academy of Engineering declared managing the nitrogen cycle one of its Grand Challenges in 2008, Preiner and his co-founders Nick Koshnick, MS '04, PhD '09, and Justin White, MS '08, PhD, '11, responded to the call.
The company's No-Wait Nitrate system, which received $2.05 million from Khosla Ventures in 2009 and has since raised $17 million from VCs led by Andreessen Horowitz, uses a field-deployed measurement tool and cloud-based data analysis to provide timely and accurate soil information. "It's extremely important that soils today are tested to the proper depth and tested for the nutrients that the plant roots will take up," says Solum customer Tim Smith, an agronomist who advises growers.
That the seeds of both companies took root on the Farm is no accident, given its proximity to some of the nation's most productive agricultural regions. "We were looking at three things at once," says Preiner, "the changing agricultural market which was facing regulatory pressure around nitrogen use, what a technological solution might be, and also how to build a successful company. We couldn't have founded [Solum] without Stanford."
Julie Muller Mitchell, '79, is a writer in San Francisco.
