The potential of digital agriculture to contribute to higher yields through multiple images and high-tech calculations by machinery is slowly being realized, but farmers can’t just sit back and let the machines do the job, panelists at the American Seed Trade Association conference in Chicago said Monday.
Producers need to interact with the technology in order to realize its benefits down the road, said Richard Marsh, global digital agriculture lead for Farmers Edge, a precision ag company with headquarters in Winnipeg, Canada.
Marsh spoke about machine learning, the process by machines can take data and “learn,” in the case of agriculture, what plants need.
But machine learning is “just about impossible if we don’t have good data,” Marsh said. Without farmers verifying and labeling what they’ve done in a field, “it’s really not of great use to us in machine learning.”
In other words, every time they go into a field, farmers need to input all the right data into their apps: what they sprayed and how much, for example.
Showing a slide of a nonproductive area of a canola field among otherwise healthy canola, he said that without knowing the farmer had performed a “double pass” with herbicides, the machine doesn’t have a chance of learning what to do next time.
Acknowledging that recording one’s every action of observation in a field can be “a pain,” Marsh said it’s nonetheless very important for generating a valid set of data over a period of years. “With enough calibrated data, we can use machine learning” to predict disease and yields, he said.
“We really want people to label what they’re doing,” he said. “It’s just a change of habit.”
Drones, which burst onto the ag scene a few years ago, are one tool to gather the information necessary, but their promise has been oversold in many cases, said Jim Love, light robotics manager for Beck’s Hybrids in Indiana.
He recalled hearing a presentation from a company that promised, without supporting data, that use of its drone would add $10/acre to growers’ bottom lines.
“This thing is a fine-tuning tool,” he said, adding that he has, in fact, recommended in many cases that customers not buy drones, or at least scale back their desire to buy a new or more expensive model when what they have works just fine.
“If you undersell and overdeliver, people are always more happy,” Love said. “The hardware is affordable” and the price can be inexpensive when looked at the cost per acre, he said.
But you cannot leave out the human element, Love said. Until a grower knows what he’s looking at, there’s no point in just sending him a lot of data.
“Almost all this digital stuff requires some type of human interaction,” Love said. “All of this data … needs somebody to help it along. It needs some babysitting once you get it into the field.”
Another speaker, EarthSense CEO Chinmay Soman, said “Data is no good unless people put in a little bit of time teaching the computers what to make of it.”
EarthSense has been testing a small, autonomous robot called TerraSentia that can navigate fields under the crop canopy and collect data such as plant height, stand-counts and stem width. In June, the company raised $1 million for research to help it commercialize the field phenotyping platform.
“The robot goes through the field, uploads the data, then you can download a CSV (file) for the whole field,” he said.
Soman said digital ag has another, perhaps unintended, benefit: It’s cool to use.
“It’s about attracting the next generation of talent,” he said. “People who wouldn’t have considered ag as a career, they’re thinking — I get to play with robots, I get to play with drones.”
For more news, go to www.Agri-Pulse.com.
