Listen to this article
|
I was fortunate to be invited to attend this event on behalf of The Robot Report. The event featured a full day of in-depth presentations and in-field demonstrations at the newly christened John Deere test farm near Elgin TX, northeast of Austin.
John Deere is a software company
John Deere CTO Jahmy Hindman kicked off the event with a high-level overview of technology development at John Deere (JD). Over the last five years, the company has grown its software development group by 350% and now employs 4,300 software engineers.
“Our mission by 2030 is to make sure every one of the 10 trillion corn and soybean seeds can be planted, cared for, and harvested autonomously. If the farmer chooses to do so, that’s going to require this full complement of the John Deere tech stack”, said Hindman.
Deanna Kovar, John Deere’s VP of Production and Precision Ag Systems, took the crowd down history lane by recapping some of the key milestones for John Deere precision agriculture since its founding in 1837.
Autosteer is the foundation
The first key high technology milestone for John Deere was the introduction of GPS – Navcom in the 1990s. This enabled JD to deploy autosteering for tractors while improving the accuracy and repeatability of various crop management processes. Autosteering also helps to reduce the mental stress for the tractor driver, while also reducing the skill level necessary to deliver precision agriculture. For the farmer, this enables less experienced drivers (like teenage children/grandchildren or farmhands) to drive the tractor with accurate results.
Autosteer is now an option on every modern John Deere tractor. During the harvest, grain wagons can sync with the combine and match the speed and direction of the combine during the grain offloading process. This simple process reduces the mental stress and fatigue on the combine driver and helps with the throughput during the time-critical harvesting process.
The next big technology milestone came in early 2010 when JD introduced IOT infrastructure to the farm that connected every machine and farm implement. This was the start of the precision agriculture data tsunami and required additional innovation to ingest and process a growing mountain of data. This information would form the basis of knowledge and decision-making tools like yield maps, stress maps and chemical application maps.
Autonomy makes its debut in 2022
Finally, in 2022, JD introduced its first fully autonomous tractor, the John Deere 8R, which leverages computer vision and machine learning to enable the tractor to perceive the world around it, identify obstacles and safely operate without an onboard operator.
Along the way, JD acquired Blue River Technology and Bear Flag Robotics to help infuse the JD engineering team with the IP and technical resources necessary to grow at the pace necessary to remain a leading AgTech supplier. John Deere also maintains a Silicon Valley development center and laboratory where it supports the accelerating push for innovation.
Tractor autonomy remains a key investment area for JD. However, there are other key areas of innovation that are viable for agriculture automation today, including the planting and spraying workflows.
AI is key to smart spraying
Another recent innovation is the See and Spray technology that JD acquired through Blue River Technology. JD has commercialized this technology that leverages machine vision and AI/ML to distinguish and identify crop plants from weeds and then only spray the parts of the field when a weed plant is present. Note that this workflow is highly competitive with a number of competitors in this space using everything from lasers, to robotic weeding implements, to vision-guided chemical applications.
The latest generation of the See and Spray solution now includes an auto-leveling carbon fiber boom along with dual nozzles that can deliver two different chemicals in a single pass.
John Deere technology helps farmers save money
In planting, the key innovation is called ExactShot and is focused on optimizing one of the most important processes in farming. Mechanized seed drills have been around for hundreds of years. Until recently, however, the classic seed drill was “open loop” where the farmer would set up the seed drill, run it for a couple of minutes and then stop the tractor, and go through the planted seed furrows to determine the success of the planting process by digging through the planted ground with a screwdriver. He would then make some mechanical adjustments, run it for a few more minutes, stop, recheck, etc. This process didn’t easily account for local soil variability between fields or even within a given field. The results were not optimized with some individual plants suffering from inadequate planting depth.
With ExactEmerge row units and ExactShot technology, John Deere is using sensors, vision cameras, machine learning and onboard processing to completely optimize the planting process. The ExactEmerge system can see the profile of the furrow in real-time and make adjustments to the discs to create a perfect seedbed for every seed, even as topsoil conditions change. ExactEmerge is capable of planting over 40 seeds per second, 12 hours per day.
The ExactShot technology (see video below) sprays a shot of fertilizer on each seed, optimizing fertilizer usage. Older generations of technology would simply spray a constant amount of fertilizer throughout the process, wasting any fertilizer not immediately next to a seed. Up to 54 ExactShot planting Rowbots can be configured in parallel on a modern planting platform.
Tillage is the first workflow to be fully automated by the new John Deere 8R autonomous tractor. There are a couple of key reasons why tillage is a candidate for automation. First of all, tillage happens immediately after harvest, a time when the farmer is often busy getting the harvest to market.
Secondly, the tillage implement is “dumb.” It simply needs to be dragged through the soil and the accuracy is less critical than with planting or harvesting. Also, unlike planting or harvesting, there is nothing to reload or offload from the tractor/implement. This means that the autonomous tractor can run until it needs to be refueled, which can be up to 8 hours. Lastly, tillage is a process that is often done by farm hands or the next generation (kids or grandkids), and this is a key area of labor shortage for many farmers today.
The net result is that tillage is a key target market for autonomous tractors, and John Deere is deploying a large number of autonomous tractors into the field this upcoming growing season. As autonomy improves and becomes more robust, autonomous tractors will be deployed into other workflows in future years.
In March 2023, John Deere acquired SparkAI, a company that developed a “human in the loop” technology and process to help recover autonomous vehicles when they became confused or encountered an edge case in obstacle detection/perception. JD has been working with SparkAI for the last couple of years and will now fully employ this technology to help improve customer experience with the tractors in the field.
Rural high-speed bandwidth is a barrier to autonomy
High-speed connectivity is the one huge hurdle that remains to the successful deployment of autonomous tractors. John Deere 8R autonomous tractors can only operate when there is a high-speed, high-bandwidth connection to the tractor and to the farmer. This is required primarily to connect the tractor to the farmer who can remotely control the tractor operation via a cellphone or tablet. The second requirement is for the human-in-the-loop (SparkAI) functionality, which is the first line of support for JD.
Farms or fields which lack a high-speed cellular connection can not currently operate an autonomous 8R tractor. Brazil is a key agricultural market for John Deere and it, unfortunately, is also one of the countries with a low percentage of rural high-speed connectivity coverage. Thus, Brazil is likely to be locked out of using autonomous tractors until the connectivity issue is resolved.
John Deere is actively exploring solutions to this problem, including the potential partnership with low-earth orbiting, high-speed bandwidth connecting satellite networks. I would predict either a deep partnership or acquisition of one of these companies in the near future, as this is critical to John Deere’s future autonomous roadmap.
John Deere test farm
On the second day of the event, John Deere hosted the media at its new Austin test farm about 40 miles northeast of downtown. On the farm, we were allowed to see, touch and ride on all of the equipment in operation at several demo stations around the property. The company purchased the property in Texas so that testing of new technology can happen 12 months out of the year, and keep pace with the accelerating technology innovation curve in AgTech.
Check out the latest episode of The Robot Report Podcast where Steve Crowe and I recap the JD Tech Summit and talk to Jorge Heraud, VP of Autonomay at JD:
Tell Us What You Think!