Glacier FarmMedia — Spot spraying isn’t a new idea, and the equipment capable of selective application has been around for decades — but the technology that supports this approach is advancing.
Tom Wolf, an application specialist with Agrimetrix Research and Training and Sprayers101, recently shared some of the new research, technology developments and application techniques during the Western Canadian Crop Production Show in Saskatoon.
The biggest jump has been video processors and graphic processing units — the technology that decides whether a plant is a weed, identifies its size and chooses how many nozzles to activate.
Read Also
Machinery automation runs through 2025 Agritechnica innovation awards
Computer vision and AI processing for farm machinery show up many times in Agritechnica’s 2025 innovation award winners.
“These processors are basically doing calculations where there are hundreds of millions of parameters that they have to solve for in relation to the image that they’ve just seen,” Wolf says.
“They have to do this in 150 milliseconds, because they look about a metre or two ahead. By the time the boom gets there, they have to make a decision.”
For this level of the technology to work, though, someone has to teach the algorithm how to do this.
Wolf says his colleague Guy Coleman from Copenhagen takes an image of a field about a square metre in size, then manually selects the weeds in the image and outlines their leaves and stems.
This tedious task is essential to teach the system how to properly identify a weed; the process is repeated 10,000 times per weed.
Image processing with algorithms is “full computation,” Wolf says. The system uses an RGB camera and needs no help from the farmer if he has the algorithm.
This method is one of three ways to perform spot spraying; the other two are reflectance-based, and an RGB camera that detects rows.
Unlike with traditional spray systems, though, it’s not just the machine and hardware a producer will need to buy. Producers will likely also have to pay for access to the software and algorithms to run the spot spray systems, Wolf says.
Companies such as Bilberry and John Deere charge yearly fees to access the necessary software for equipment, he says. Bilberry is running $4 per acre, per year, and John Deere charges for the acres not sprayed.
“You also have to consider cost, like what if you miss? What are the consequences of a miss that we didn’t anticipate?” Wolf says, posing hypotheticals to the audience. “Do you have to spray again? Do you have a yield loss? Do you have a harvestability issue?”
All factors considered, there are savings for the producer, but not as much as one may think.
Wolf’s advice is to take the money saved from spot spraying and further invest it in weed control.
Consistent boom height is absolutely critical for spot spraying, he says.
“If the boom is too high or too low, the camera isn’t sure that it sees what it’s supposed to see and it goes into what we call ‘fallback.’
“So if your boom sways up, all the nozzles go on. It (the camera) says, ‘I can’t see. I’m going to make sure everything gets covered.’”
Height and speed are important to reduce sway. They also ensure proper concentration of application and correct overlap for multiple nozzles, or if switching to broadcast.
Some systems use an overlap mode, in which it turns on the target nozzle but also the neighbouring nozzles.
“We’re no longer spot spraying with a band, but rather a broadcast — a mini-broadcast screen. Now your overlap issue is more or less gone. On the downside, you’ve used one and a half times more product than you really need to,” Wolf says.
A spot spraying system can be expensive, especially one that runs on software and algorithms, but it can serve additional purposes, he says. Spot sprayers can be used for herbicides, but also to desiccate crops more accurately, map weed species and create a general weed map.
Wolf says there is economic power in weed maps because they help agronomic decision making for farmers and agronomists alike.
Companies such as SWAT Maps have cameras that can sit atop sprayer booms and then use an algorithm to map weeds. “From an agronomist perspective, this is pure gold,” Wolf says.
Weed maps are also useful to companies and agronomists to provide further analysis to a farmer. It can offer insight to crop and weed emergence, best application methods and more clarity in target areas.
