Embracing IoT tech, John Deere plows a path to autonomous machines

When you think about it, John Deere’s self-propelled agricultural spraying machines are really just big IoT devices. But big as the products may be, many of the overall design issues and considerations are the same as what every other developer of IoT devices is confronted by—they’re just different in the agricultural space.

Joel Hergenreter, manager of tech stack system test and validation at John Deere, sat down to talk to FierceElectronics about the IoT based technologies his team is working with, the importance of understanding customer needs, and some of the exciting opportunities ahead for the company.

FierceElectronics (FE): You have a pretty long job title. What do you actually do?

Joel Hergenreter: My group focuses on the agile delivery process for technology and for bringing software solutions to the market. To avoid getting caught up in focusing purely on the technology, my group’s priority is to make sure that the technology is bringing value to the customer. We make sure that we’re building the right product for the market and that it works right.

That’s where the validation part comes into play. We go through all sorts of testing, starting with in-house testing and then testing in controlled environments and then ultimately testing in the real world with real customers. We also put a lot of effort into ensuring that the technology is intuitive. If the equipment seems too complex and it doesn’t show the value immediately, then the customer will not view it as a good business investment or worth putting the time into learning it.

I feel very fortunate to have joined the company right when we were introducing things like GPS at the forefront and being along for the journey as we’ve embraced and adopted new technologies like machine learning. Each of these new innovations is directly impacting our customers, who are ready for it because they know the technology is going to continue to make them more efficient and effective.

FE: How many sensors does John Deere typically have on a machine, and how powerful are they?

Hergenreter: Our self-propelled sprayer is a good product to highlight. We have over 400 sensors on the sprayer, which automatically applies liquid herbicides on crops to keep them healthy. Some 250 of those sensors are located on the boom at the back of the machine which is up to 120 ft wide. We have been in this market for about 20 years, and the technology is good enough that we can get down to sub-inch accuracy of the application of the herbicide.  We use other vision sensors to make sure we are traveling in the right area of the field and not driving over the crop itself or causing any harm to the plants. We also rely on a telematics device to make sure all of these sensed measurements are transitioned to the cloud so a remote manager can view this data in real time. For connecting to the cloud, all of our equipment has 4G modems, which we started making standard a decade ago.

FE: Can you discuss some of your team’s design strategies for managing power, which I wonder might be less of a challenge since you have such big machines? 

Hergenreter: Just because our machines are big doesn’t mean we get a pass on power issues. We definitely have challenges—they’re just different!  For example, we are handling a lot of data. As a frame of reference, Twitter has about 6,000 tweets per second, while we are collecting about 2,000 sensor readings per second - and that is just per machine. On a typical day we record 10-15 million sensor measurements per second across all of our machines! So even though we have big machines we still require a lot of extra power, which also generates heat. With so many modems and controllers on the machine, we have to be able to manage the heat.

Our machines also operate in a multitude of field environments, and so we have to make sure that the machines are ruggedized. We sell the same core product in over 100 different countries and we have to design for every type of climate you could imagine.

We employ multiple controllers on each machine, whose job it is to just make sure that the equipment is staying connected. Keeping machines in sync while in the field is a critical task that is kind of a unique variant of the ag market. We have products where one machine is controlling another machine when they are driving through the field. So, you have a couple of multi-hundred-thousand-dollar pieces of equipment operating within inches of each other through rough terrain. We need our technologies working perfectly to do their job and to keep the operator safe.

FE: How are IoT-based technologies working in tandem with other technologies  (such as robotics, sensors, 5G) to enable real-time data sharing, machine-to-machine communication and more efficient practices on the farm? 

Hergenreter: First, we start with customers that have a problem. We try to understand the problem very deeply to ensure that the technologies that we integrate will solve that problem. We do this by meeting with customers to go over new designs early in the process to make sure we are going down the right path. Then we go into virtual engineering verification to ensure that the  products that we are developing and all of the inherent technologies—whether it is machine learning or robotics or 5G—are working correctly in the virtual environment. Then we take the machine out to a controlled environment to ensure it works there. Only at that point do we take the product out into the wild to our customers.

FE: What can other industries learn from John Deere about IoT-based technology?

Hergenreter: My advice would be to remember that it all comes back to your customers and deeply understanding their problems and developing the right solutions to address them. For example, our customers have a  short window of opportunity—just 5 to 10 days—to plant their crops. Should they miss their opening for weather or whatever that could turn into not planting the crop at all. Once we gain an understanding of the true issues they’re facing, we can go about collecting the right data that gives them actionable information. For example, if we know a rainstorm is coming, that could mean making the decision to increase the speed of the machines to get more seeds planted in advance of the impending bad weather.

FE: How does the cloud and expert alerts play a role in updating software/firmware on machines in the field? 

Hergenreter: Back in 2010 when we first released machines that had connectivity, we didn’t realize how valuable that would turn out to be in updating our software. We  have a 4G modem on every machine and we manage data communications over a cellular network. We are releasing new features three times a year on the embedded side and six times a year on the digital side, and doing it over the air allows us to continue to update our entire fleet every several months. That allows us to provide new value to the customers on a continual basis.

FE: What do you view as the advantages of 5G for John Deere’s technology?

Hergenreter: Of course, there is the benefit of lower latency and faster speed to the cloud. But the major advantage of 5G for us is that it will allow remote farm managers to see what’s going in real time from the perspective of what one machine is doing relative to another. Also, we currently manage the land in approximately square-meter-size plots. With 5G, we can start to break that plot down into individual ears of corn, for example, ensuring that we are getting the right nutrients to the right plant. By doing so we can start to build more trust with the remote manager that the machine is doing the right thing and this is going to be a big step toward truly autonomous equipment.

Editor’s Note: John Deere’s Joel Hergenreter will be speaking on IoT and Wireless technologies at Sensors Innovation Week, a digital event series taking place November 16-18, 2020. For more information and to register for your free pass, click here.


How John Deere got good at AI

Deere shows off smart weed control machine that relies on AI with Nvidia chip