Stout: the Best AgTech Comes from Growers
Born inside of a large grower-shipper operation, Stout takes a grower-centric approach to artificial intelligence.
Sponsorisé - Cet article contient du contenu sponsorisé.
In Plato’s “Republic,” he famously wrote that “our need will be the real creator.” Few industries understand this better than agriculture, where challenges from labor shortages to unpredictable weather conditions create the need for more reliable, less expensive, and more effective solutions.
This was the case when a large grower-shipper in California’s Salinas Valley found its weed problem spiraling out of control. Weeding crews were increasingly difficult to find, highly unpredictable and more expensive every year. Chemical herbicides were heavily regulated, expensive and hazardous for employees. Mechanical weeders had become increasingly popular, but often eliminated crops along with the weeds.
Farming operations, especially those with significant acreage, needed something better. So, the first Smart Cultivator was built. It had three goals to achieve; it had to be super reliable (aka extremely durable), easy to operate in the field and highly accurate (no crop kills). This required a new approach from everything in the market.
For a grower, reliability means it must operate day or night, in all field conditions and in all types of weather. It must be built to take a beating in the field, because farm implements get knocked around. And it must be easy to troubleshoot and fix. Fragile machines have no place in the field.
One look at the Smart Cultivator and it’s easy to see how this goal was achieved. The machine is built like a tank, featuring aircraft-quality billeted aluminum and stainless steel, waterproof components and insightful mechanical elements like shock absorbers that enable the machine to operate in rocky soil without breaking the cultivation arms.
Stout’s oldest machines have over 6,000 hours on them and are on their second tractor. Some of Stout’s customers run their Smart Cultivators 18 hours a day, six days a week, simply changing shifts on the tractor.
Making the machine easy to operate in the field required a visual-based approach to information. Many operators don’t speak English, so the Smart Cultivator uses a simple touchscreen control panel. Connections on the machine are outfitted with LEDs to enable quick visual inspection and troubleshooting. The Smart Cultivator connects to the tractor with only the 3-point hitch and PTO. Every aspect of the machine is designed for in-field operational simplicity.
“If you need an engineer to operate or troubleshoot the machine, it’s not built by a grower,” says Stout CEO Brent Shedd. “Growers need reliable machines that are easy to operate and get the job done.”
The reason many mechanical weeders available today have trouble killing crops is because they’re not using artificial intelligence. You hear terms like “green on green” or “green on brown” associated with machine vision, but that simply means the vision system is acting based on blobs of color. It can’t differentiate between a crop or a weed.
To eliminate crop kills, Stout had to develop its own proprietary Stout True Vision system, the core of which is a convolutional neural network and monolithic artificial intelligence model which identifies each weed and each crop in the field. The system prioritizes the crop above all else – just as a grower would. If a weed is too close to take out without harming the crop, the Smart Cultivator defaults to leaving the weed rather than harming the crop.
Deep Learning systems like Stout’s are difficult and expensive to build, which is why most automated weeders stick to the green-on-green or green-on-brown blob detection, even if it means killing crops along with weeds.
“If you want to know whether a system actually uses artificial intelligence, just ask what their training protocol is,” says Shedd. “AI is like a child; it must be trained to identify things. That training requires large amounts of data, imagery, and effort.”
Stout’s training protocol requires the AI system correctly identify a commodity at 99.9 percent accuracy before the commodity is added to the list that the Smart Cultivator operates in. That list currently includes over 24 commodities, but the machines have the capacity to work on any crop that can be mechanically cultivated.
Growers are very ROI-focused which led to another key innovation built into the Smart Cultivator. Stout’s machines are software-defined, which means that the Smart Cultivator’s capabilities can be increased via over-the-air software updates. No mechanical changes necessary. “A Stout machine will continue to add more and more value to a grower’s program the longer they own it,” states Shedd. Something very few farming tools can claim.
Service and support are also critical to growers, which is why Stout machines are globally distributed by New Holland Agriculture dealers. With 1,799 dealers operating 4,031 locations worldwide, Stout customers always have the service and support they need nearby. Today, Stout has machines operating throughout the US, UK, Europe, Mexico and, later this year, in Australia.
AgTech is growing exponentially, but few products achieve significant farmer adoption because they lack the insights that derive from industry knowledge.
“Stout’s superpower has always been our grower DNA,” says Shedd. “Growers are famously innovative and having that insight, those resources and that focus ensures that Stout’s products will continue to define what really works in AgTech for the foreseeable future.”