AI-trained machines slash labor costs

Postdoctoral researcher Connel Ching’anda stands by a machine that uses steam to disinfect soil of pathogens that can harm leafy green crops. Steve Fennimore and his team, of the UC Davis Department of Plant Sciences, tested the equipment last year in Salinas fields, and they demonstrated its use during Automated Technology Field Day, also in Salinas, in June. (Photo courtesy Jeffrey Mitchell/UC Davis)

Experimental robots are reducing the costs of hand-weeding by learning the difference between weeds and lettuce. In addition, steam can clear the soil of fungi and spores that cause lettuce and spinach to wilt, reducing the need for chemical herbicides in the bargain, according to the latest research by Steve Fennimore and his lab at the UC Davis Department of Plant Sciences.

Fennimore, a professor of Cooperative Extension, presented his findings at a recent meeting of the California Leafy Greens Research Board, a body of the state Department of Food and Agriculture, which is helping to fund the work. It was attended by growers, association representatives and educators. This research would help them solve three big problems:

“What everyone is facing is higher costs, fewer personnel and greater difficulty using pesticides due to regulations,” said Fennimore, who also is a specialist with UC Cooperative Extension.

He and team members Richard Smith (UCCE Farm Advisor) and Nelly Guerra (UC Davis Fennimore Lab graduate student) evaluated automated weeders that were trained using artificial intelligence for use in lettuce fields. In tests last year near Salinas, Calif., smart weeders built by FarmWise Labs Inc. and Stout Industrial Technology, Inc., removed between 32 percent and nearly 100 percent of purslane and other weeds, Fennimore said. That reduced the need for hand-weeding between 13 and 62 percent. The weeders proved more cost-effective in fields where there were more weeds, Fennimore added.

The team also tested steam to clear soil of diseases that cause leafy greens to wilt and turn brown at the edges. During trials last summer in the Salinas Valley and in Yuma, Ariz., team members measured soil pathogens before and after steaming the rows where seed is planted. They found significant reduction in the fungus fusarium; in the tiny balls, or microsclerotia, that allow fungus to survive in the soil; and in pythium, a water-born mold. Steam treatments also boosted lettuce yield and reduced weeds. The team collaborated with Mark Siemens, of the University of Arizona.

Building on those successes, Fennimore and team this year are combining steam with standard cultivation to see if they can control weeds 100 percent and reduce hand-weeding to zero.

Media resource: TrinaKleist, tkleist@ucdavis.edu, (530) 754-6148 or (530) 601-6846