Exploring Ag Robotics’ Potential – and Challenges

In a clear sign that agriculture’s continuing adoption of robotics will have human as well as technological and logistical ramifications, protesters from France’s Yellow Vest movement unexpectedly seized the stage of the third International Forum of Agricultural Robotics (FIRA) in Toulouse, France in December and held it for about 45 minutes until organizers and attendees shooed them off and resumed the conference.

If robotics’ potential to impact agriculture has attracted the attention of societal activists, it’s clearly done much the same among FIRA attendees, who represented both startups and larger tech and equipment companies as well as growers, consultants, financiers, suppliers of peripheral products and others with a keen interest in robotics. Many marveled that application of ag robotics has picked up steam after years of sometimes hit-and-miss development and skepticism in the wider agricultural community. Yet challenges still remain.


Robotics, artificial intelligence, and machine learning are evolving so quickly, U.S. precision consultant Marc Vanacht noted that his presentation at the conference was obsolete the moment he completed it. In addition to almost innumerable startup technologies adapted to regional and local uses, most new developments around the world still come from labs at national governments and universities, including projects reviewed by noted global robotics experts Prof. Salah Sukkarieh of the University of Sydney in Australia and Dr. Noboru Noguchi of Hokkaido University in Japan.

There also are increasing numbers of corporate-funded technologies, Vanacht noted, including John Deere’s Blue River Technology and Bosch Deepfield Robotics’ Bonirob, both of which offer high-tech weed-control solutions. Still, “the American model of venture capital has not understood yet how important ag robotics are, or don’t feel comfortable with it, don’t know how to make money with it,” Vanacht said.

So, Are Robotics Feasible for Cotton?

That’s the question that Cotton Incorporated and numerous university research partners have been working to answer for the past few years. So far, the results and possibilities are very encouraging.

“Cotton Incorporated is not going into the robotics business,” explained Dr. Ed Barnes, the organization’s Senior Director, Agricultural and Environmental Research. “What we are trying to do is get a baseline amount of data to document if this is feasible in cotton. And if it is feasible, hopefully the private sector will pick it up.”

In a presentation at the 2019 Beltwide Cotton Conferences, Barnes pointed out that many current factors in agriculture are already leading to robotic farming – things like sensor-based management, high throughput phenotyping, machine vision and AI, herbicide failures, autonomous auto-steer equipment and continued labor shortages.

“A lot of things are coming together,” he said. “It’s just unavoidable.”

Of all the current factors involved, Barnes believes that herbicide failures will be the key driver to move toward robotics. And that seems to be universal. Growers at FIRA expressed a clear preference for weeders as their “dream” robot.

Which is, of course, why all of the initial research in cotton is focused on harvesting.

“Our current harvesting system has provided labor savings and a lot of efficiencies, but we have to do things in 2,000 acre increments,” he noted. “We can’t go from 100 to 500 acres of cotton. We have to go 2,000 to 4,000 to 6,000 acres, so it’s not real scalable. And it’s a heavy machine with soil compaction risks.

“So why robotic harvesting when weeds are going to be driving this change?” he asked. “The concept we’re trying to start with is if we can harvest cotton frequently – like up to 25 times during the season – we’re reducing risk and loss from hurricanes, aphids and whiteflies. Sticky cotton becomes less of a problem. We’re going to preserve fiber quality and reduce weathering, plus create more efficiencies at the gin.”

The theory is that by harvesting open bolls frequently – even while leaves are green – all lint from a plant is accounted for and should provide more uniform fiber properties that will be of value to textile mills. The units go the field to pick the bottom crop first, with subsequent trips for the middle and top crop as those bolls open and are ready. And in terms of scalability, current studies are examining the use of one robot for every 11 acres, making exponential expansion much simpler.

In its most futuristic concept, Dr. Joe Maja, who heads up the robotics work being done at Clemson University, could see a swarm of machines in a field – all communicating with each other – while UAVs fly overhead to scout fields and determine where bolls are ready to pick.

Barnes foresees a time when these units could be solar powered to help save money. But the key thing is the units under study are multi-functional. They’re not just built for harvesting. The platform, guidance, software and other autonomous systems being used can be bought off the shelf.

And yes, now that the platform is pretty much in place, studies begin this year on ways to turn the harvester into a weeder, as well.

Various aspects of the robotics studies are being conducted under Cotton Incorporated supervision at Clemson, Georgia, Texas A&M, Tennessee, North Texas and Kansas State. Those studies range from development of the picking (or vacuum) mechanism to economic reviews to help determine if/how fiber uniformity and quality from robotic-picked lint provides increased market value.

“Folks ask me how long is this going to take,” said Barnes. “I’ve used a 10-year timeframe starting in 2018, so we’ll hopefully see this within the next decade. But even if this turns out to be a no-go, some of the things we’ve already learned are going to have benefits. So it will not be a wasted effort.”

Workerless Farms?

It’s a concerning topic in Europe, where robotic development is moving rapidly. And while the Yellow Vests’ concerns about workerless farms were widely dismissed by FIRA attendees, the “Hands-Free Farm” – a fully autonomous operation working across multiple crops and fields using swarms of autonomous precision machines and remote agronomy – is in the final planning stages by Harper Adams University in the United Kingdom.

“All of the equipment we use will be bought on eBay and Amazon to prove that this equipment is available to all,” said Jonathan Gill, a robotics engineer and Harper Adams researcher. The goal of the Hands-Free Farm will, of course, be that no human ever enters the field.

That raises the question: what will become of agronomists and other roles, whether “remote” or not? One FIRA attendee asked Gaëtan Severac, a co-founder of Naïo Technologies in France, about a common fear that artificial intelligence will replace consultants.

“You can’t go against the general flow of society,” Severac said. “I think the job of consultants will change, but they will not disappear.”

Such are the continuing societal questions robotics developers will need to answer as the sector matures. Aymeric Barthes, Severac’s partner at Naïo, perhaps put it best when he said, “Robotics has to prove its value, as tractors did in the past. It is a long and hard way, but that is the challenge that we are choosing to take up. All of us in this place – we are pioneers.”


From Cotton Grower Magazine – February 2019