Heavy equipment maker John Deere says its machines traverse one third of Earth’s surface each year. As they till, plant, fertilize and harvest, the Deere machines also collect billions of measurements that describe soil, crops and weather conditions.
The value of this data was a major theme this week at an event Deere hosted in Austin for journalists and analysts. First, the company explained how farmers make use of their own data for near real-time decision making. The connected tractors send data over LTE to the AWS cloud, where it is processed and returned to farmers’ smartphones in 20 minutes to an hour, according to Lane Arthur, VP of data, applications and analytics in Deere’s Intelligent Solutions Group.
Farmers send Deere data about their crops and soil, as well as the status of their equipment, enabling the manufacturer to help with real-time troubleshooting if a machine is not working. The ability to fix equipment fast is critical during planting windows, one farmer told the audience. “A day can make or break your entire year,” he said.
In addition, many farmers allow their data to be anonymized and aggregated, enabling Deere to create a data set from multiple sources worldwide. “This is an asset none of our competitors have,” Lane noted. He explained over lunch how he started Deere’s migration to AWS years ago, and now uses thousands of AWS servers each spring and fall, when North American farmers are planting and harvesting.
Deere has 500,000 connected machines between its construction and agricultural divisions, and is targeting 1.5 million connected by 2026. A small subset of its tractors drive autonomously, freeing farmers for other work during busy seasons. But whether they are driven by humans or not, the connected machines act as “sensors on wheels,” Deere said.
As the amount of data collected goes up, Lane expects to need more processing power closer to the tractors and other connected equipment. Although the machines already have servers beneath their seats, Lane predicts that even more compute power will be needed. “The edge is coming, and I am going to need that,” he said.
Already, a significant amount of data is processed on the tractors themselves. The company’s See and Spray technology uses smart cameras to distinguish between crops and weeds, triggering sprayers to spray herbicide on the weeds and fertilizer on the crops.
Satellite connectivity
Jonny Spendlove, senior product manager at Deere, updated the audience on Deere’s talks with satellite connectivity providers, projecting that as the company approaches its goal of 1.5 million connected machines, 10-15% of those will use satellites to connect to the cloud.
Spendlove said Deere’s satellite solution will probably involve low Earth orbit satellites (LEOs), and he said the company is targeting speeds of 5 Mbps for both the uplink and the downlink. He said Deere wants latency of 500 milliseconds or less.
“Over time we are going to be introducing use cases that are going to be requiring lower latency and higher bandwidth,” he said. “We see billions of dollars of value in unlocking these use cases that cannot exist unless the tractor is connected.”
Spendlove also noted that in areas where tractors need satellites in order to connect to the internet, people also need a way to connect. “We are finding ways to solve some of those questions and problems … that is something we are considering,” he said in answer to an analyst question.
Deere recently chose satellite startup Albedo as the winner of a competition the equipment maker held for technology innovators. Albedo plans to launch 24 very low Earth orbit Satellites (VLEOs), and promoted these as a way to take high-resolution images rather than as a way to connect machines or smartphones.