John Hart 

Ed Barnes knows the pulse of technology in cotton, from the farm to the mill.

Technology has advanced rapidly since the agricultural engineer joined Cotton Incorporated in 2002, making it an exciting time to work in the industry. Advances show no sign of slowing, making cotton more proficient every step of the way in the years ahead.

Barnes, senior director of agricultural research at Cotton Incorporated, highlights the use of robotics and machine learning that will work to improve efficiency at both the farm and the gin. On the top of his wish list is a robotic cotton harvester that will both lower costs and improve the quality of harvested cotton.

Barnes has actively promoted the adoption of multiple precision farming technologies for cotton production — working with research and Extension to educate producers on how precision farming can increase in-field efficiencies and improve bottom line. He is recognized internationally as an expert on cotton harvesting technology.

Strong demand

Beyond any measure, the John Deere 7760 revolutionized cotton harvesting. In August, the company unveiled the newest versions of the machine, the CP770 Cotton Picker and CS770 Stripper. With the exclusive round module builder, the John Deere harvesters have made the modules a familiar sight in cotton fields across the country.

Ed Barnes, in front of the Truetzschler TST5 Contamination Detection and Cleaning Line at Cotton Incorporated’s fiber processing research center in Cary, N.C. The TST5 removes plastics and foreign material from cotton fiber at the mill. 

“The John Deere 7760, whether a stripper or a picker, is a great machine. It reduced labor requirements at harvest which is why it really overtook the market, and its data capability is really exciting. That said, it’s a $1 million machine, list price. Our economic analysis, in round numbers, suggests that for the picker version, you’re going to have to run 2,000 acres through that machine for a return on your investment,” Barnes said in an interview with Southeast Farm Press at his office at Cotton Incorporated headquarters in Cary, N.C.

Barnes acknowledges that many farmers do indeed produce more than 2,000 acres of cotton, which is a big chunk of land, but many farmers, particularly in the Southeast, are rotating cotton with other crops and may just have 500 to 600 acres of cotton in a given year. “There’s not really a harvester they can own for that amount of acreage.”

Barnes said there is strong demand for a harvesting system that does not require a $1 million cotton picker. Cotton Incorporated is working with the University of Georgia, Clemson University, Mississippi State University, and other groups in developing an automated system that would not only be an alternative to a big harvester, but could make multiple passes through the field.

Both the University of Georgia and Clemson University have prototypes of robotic cotton harvesters that are being used in research settings. The UGA machine is called “Red Rover” while the Clemson robotic harvester is called “Husky Harvester.” Glenn Rains is leading the UGA research while Joe Maja is directing the Clemson project.

Barnes said the beauty of robotic harvesters is they can harvest cotton as soon as the bolls open, which will mean better fiber quality and color since the cotton can be picked before it is rained on. The challenge with machine harvesting is that the first boll opens and won’t be harvested in 50 days, and at the bottom of the plant, these bolls can be lost to boll rot.

Not pie in the sky

“I think we can do this, I can’t promise we can do this. It’s not pie in the sky. It’s a small harvester, maybe they’re one row, maybe they’re four rows. They can make multiple passes. They can harvest the bolls as they open using machine vision,” Barnes said.

“We started this project three years ago. When we started, I thought we would have a prototype that would be feasible but not for sale in 10 years. We still have seven years to go. We’ve made a lot of progress,” he added.

These new robotic harvesters clearly won’t be able to harvest as much cotton as one machine harvester. But Barnes noted that the robotic harvester is expected to have a price point of $20,000 per machine, compared to $1 million for a baler picker. In other words, you could buy 50 of the robotic pickers for the price of one baler picker.

Certainly, cotton farmers will need more than just one robotic picker and each picker will need to make multiple passes to be effective, but at just $20,000 per unit, a farmer could produce well under 2,000 acres and still produce cotton. “We could get to the place where it would take 50 of those machines to replace one John Deere harvester,” Barnes said.

Clemson’s Husker Harvester also uses machine vision but has a different design than the Red Rover prototype, using headers rather than a robotic arm It’s also expected to be available in the $20,000 price range.

“The other important thing about this robotic harvester is it wouldn’t just harvest cotton. It would also do weed control; it would also do pest scouting, and possibly gap planting. It becomes a utility tractor. That $1 million cotton harvester is only good for picking cotton. That’s all you can do with it. We really see a lot of reasons to push for innovation in this area. It’s a perfect solution for some farmers, no doubt about it,” Barnes said.

Getting to gin

And as far as getting cotton from the robot to the gin, Barnes said a group at Kansas State University is working on the logistics.

“I had suggested to our board something similar to the old boll buggy concept and we could return to automated modules and I got a very strong ‘No’ So for now, one scenario is we could build something very similar to what is produced by the John Deere CP770 on the edge of field and everything from the field to the gin remains the same. The idea is the baler would also be autonomous and patrol the end of the field and service several harvesters at once,” Barnes explained.

“I imagine we could see autonomous trucks in the future to transport from the field to the gin if the world goes that way.  That really becomes a societal issue if we are willing to accept full autonomous trucks on the road. With the current shortage of truck drivers, society may become accepting of the concept,” Barnes said.

As for the current system with round modules, Barnes believes we could see an autonomous system to stage modules in less than five years. He explained that a project at Texas A&M University is analyzing Global Positioning System (GPS) from round module staging in both the field and gin to try to develop some optimization routines that could be used to train drivers now and could be used to develop algorithms for an automated system in the future.

When it comes to technology at the gin, Barnes emphasized that machine learning will become even more important as gins work to improve efficiency and supply better quality bales to textile mills. He highlighted a new project where Cotton Incorporated is working with the SAS Institute, a neighbor company to Cotton Incorporated, also located in Cary. SAS is a multi-national developer of analytics software.

Barnes said the project began this year with just two gins involved. The goal is to get 10 gins involved in 2022. The goal of the project is to see how ginning affects fiber quality. The goal is to use machine learning to improve both ginning rate and fiber quality at the gin.

“The goal is to get cotton at that optimum moisture content, which is around 6%. If the cotton is too wet, you will see machine failure at the gin. If the cotton is too dry, it becomes brittle and it breaks,’ Barnes said.

In its first year, the study shows that the most important thing in deterring moisture of cotton at the gin is the field it came from. That result is likely a combination of the variety and how the cotton was handled and managed on the farm. “Ginners say some varieties do go through the gin faster than other varieties.”

More years of research are needed in this gin study, but Barnes is confident results will be produced that will make gins more efficient. Barnes said this is the first time Cotton Incorporated has worked with SAS and they are a valuable partner. 

“We’re working with John Gottula, director of crop science at SAS. John is a great partner. He has a background in cotton breeding and went to Texas Tech. He knew cotton already,” he said.