You are the co-founder of Blue River, a startup in intelligent agriculture machinery that was acquired by John Deere in 2017 – what drives you in this business and what was the rationale behind the acquisition?
My grandparents were farmers in Peru and my father had a tech startup, so I would spend half of my summer picking tomatoes and the other half learning about automation. After working for 15 years as the head of engineering for agriculture at Trimble Navigation, I started to look for a greater purpose and was drawn to emerging technology like machine learning and computer vision, which I thought could have a tremendous impact on agriculture machinery. I joined efforts with my co-founder, and focused on Blue River’s first mission to spray less pesticides through automation. John Deere has always been at the forefront of innovation so when they approached us in 2017, we were excited for the acquisition and the ability to expand our technologies together. Nearly 200 years ago, John Deere himself invented the steel plow. Fast forward 100 years and the company innovated to introduce the first tractors. Today, we continue to embrace new technology with autonomy and automation.
What is the current state of John Deere's autonomous pipeline and how do you extract and put together all the data that is gathered on the fields?
Autonomy to us is about making driverless machines. We started by automating tillage, one of the easiest operations on a farm where the farmer is preparing the ground for planting by simply driving back and forth, covering the entire field. We took already existing technologies, like to automatically steer the vehicle and added six pairs of cameras to create a 360-degree view to detect obstacles. All the data processing (inference) is done in the machine through a rugged box powered by fast Nvidia processors and GPUs. We use onboard processing instead of the cloud as these decisions need to be made in milliseconds and uploading all the imagery the machine captures in the field would take longer than that.
We collect images in three different ways: through the autonomous tractor itself, with human operators that photograph in special conditions (e.g., at night) and through the regular tractors that are driven by farmers on a daily basis. We tag all the images that present valuable details, and send them to our central hub. There a curator selects which pictures are a good learning base to generate a model, which we later deploy to all the machines with the help of a 4G modem. With a deep neural network, we can also better understand exactly what is seen by the cameras.
As we expand into operations like planting, we will capture even more images relative to that job.
In what practical ways can automation and autonomy help farmers work smarter and more sustainably?
We are living through a big revolution that is going to have a tremendous impact on farming. The tools that we have allow farmers to be more productive, profitable and sustainable in their operations. Typically, tillage gets done at the same time as harvesting so there is a peak of labor when farm employees work 18 hours per day over the course of three to four weeks. Autonomy means farmers do not have to be in the cab of their machines, which means they can focus on other value added tasks and get some much-needed hours of rest.
The concern regarding the life span and cost of batteries in electric machines is shared by many farmers. What is John Deere's stance on this issue?
Battery electric vehicle technology is an area of investment for Deere, but the size and scale of farming equipment is not the perfect use case for it. Our autonomous tractor runs continuously for 14 hours at 75% of peak power at about 230kw, the energy equivalent of 38 Tesla Model 3 Long Range batteries. Adding a battery would double the weight and size of the tractor today at four times the existing cost of the tractor in order to produce the energy required. The changes would also lead to negative agronomic impacts on the farm, like excessive soil compaction. When it comes to electrification, our focus is on small to midsize machines and we continue to look at biofuels like ethanol and renewable diesel for the farm.
How are you picturing the future of the company, especially from a people management point of view?
Our team of engineers are passionate about our mission to feed the world. As a team, we empower each other to be agile, nimble and think differently. As we look towards the future of automation and autonomy, our goal remains the same – to help our customers be more productive and profitable.
