The advice passed around goes like this: put cameras on the boom, spray only the weeds, and the chemical bill falls by most of what you used to spend. On a Kenyan block, that advice fails at the first step. The machine it describes is a 120-foot boom, a tractor and a field long enough to drive it. The best-documented trial of camera-guided spot spraying was run in Arkansas soybeans. The saving was real, but it came in at 43 to 59 percent, not most of it, and it applied to one pass in the programme. This piece goes back to what the claim was built on.
Measured by NuaSense weather stations and soil probes on Kenyan farms, over the period stated with each figure. Past readings, not a forecast.
What the Arkansas trial actually measured
The University of Arkansas research on see-and-spray systems ran for three years in soybeans. It set camera-guided spraying against broadcast applications. Post-emergence herbicide use fell by 43 to 59 percent. Season-long herbicide cost savings were 13 to 80 percent.
Look at the gap between those two ranges. Chemical use fell by roughly half, but cost savings stretched from a small saving to a very large one. A grower who hears "saves 80 percent" has picked the best end of a spread and ignored the other end.
The system uses machine learning on real-time imagery from cameras mounted on 120-foot booms. It selectively sprays weeds in row crops. The idea began as a way to thin lettuce fields to an ideal plant population, and was later turned on weeds. That history matters a little: recognition is the hard part. The spraying hardware is ordinary nozzles switched on and off.
The trial is American evidence, from large mechanised fields. Nothing in it was measured in Kenya. No Kenyan price appears in any source we hold.
Three parts, whatever the brand
A review of precision chemical weed management strategies indexed by FAO's AGRIS found that all commercial and prototype systems share three parts. Sensors and classifiers tell weed from crop. Decision algorithms sit on a controller. A precise sprayer has boom section or single-nozzle control.
That list is a useful buying filter. If a seller cannot say which of the three they are selling, you may be paying for a controller with no detection, or detection with no way to switch nozzles. The review also describes an ISOBUS protocol proposed so a user can pair different sensors with different sprayers. That tells you the market was not standardised at the time of writing.
The same review notes that farmers ask for robust systems that need little maintenance and work across different crops. On a Kenyan farm, dust, vibration and repair access are all concerns, so that request is not a footnote. A camera that needs a specialist to recalibrate is a different purchase from one a farm mechanic can service.
The review adds that precision control can come from sprayers mounted on a tractor or from autonomous robots. Neither one is a knapsack. Most spraying in Kenya still happens with a knapsack.
Green-on-brown and green-on-green are different jobs
The AGRIS review splits detection into two types. Green-on-brown sensing tells green vegetation apart from soil and crop residue. It uses visible and near-infrared wavebands. This sensing can turn single nozzles on and off in real time before sowing, after conservation tillage, and in the inter-row area. The job is straightforward: anything green on bare ground is a target.
Green-on-green is tougher. It sorts out weeds inside a standing crop. That needs more advanced imaging, like convolutional neural networks. This is what the Arkansas trial was working on: weeds among soybean rows.
In practice, that means a difference in price and reliability. A simpler green-on-brown unit fits a bare-ground pass. A green-on-green unit must tell crop from weed. The review says that takes heavier computing. If a seller quotes one headline saving, ask which of the two produced that figure.
Another option needs no camera in the crop. The review describes noting the position of each crop at sowing. Then you target herbicide at single weeds or patches once an economic threshold is exceeded. It swaps detection hardware for sowing records.
The sensitivity dial: more saving is not more profit
The most useful result from the Arkansas work is about a setting. When run at low sensitivity, the system cut herbicide use. But it raised the pigweed population by 280 percent each year. That low setting also gave a slight yield increase of about 4 bushels per acre. Researchers linked this to less herbicide injury to the crop.
So the setting that saved the most chemical created the biggest future weed bill. The researchers estimated that improper use could cost Arkansas producers upwards of $60 million per year. That estimate used an average post-emergence herbicide cost of $20 per acre.
The high sensitivity setting acted differently. With a $25,000 upgrade over a comparable sprayer, it saved $30.49 per acre across the season. It posed no increased risk compared with broadcast.
Read this as a warning about the claim. A vendor who advertises the largest percentage cut may be showing you the low-sensitivity result. A buyer should ask what the weed population looked like the following season, as well as the litres saved this one. The lowest chemical bill and the best farm are not the same thing.
The payback sums, and why Kenyan blocks fall short
The Arkansas article works out a breakeven. If you buy the $25,000 upgrade, the sprayer could pay for itself after 819 acres. For a $75,000 upgrade, that figure is 2,460 acres. Those are US dollars and US acres, and they apply to soybean growers.
Two points follow, no invention needed. First, payback depends on area treated, not on how many years you own the machine. Second, the acreage figure climbs with the price of the kit: an upgrade three times dearer needed roughly three times the acres, since 2,460 divided by 819 is about 3.0.
We have no Kenyan price for camera spot-spraying hardware and no Kenyan acreage figure, so we will not make up a breakeven for you. Instead, take your own quote, divide it by your own saving per acre, and see how many seasons your own area needs. If your block is well under those American acreages, the numbers are unlikely to close on owned hardware.
For many Kenyan farms, the honest answer is no purchase. If a contractor offers the service, hiring it shifts the capital cost to someone with enough acres to justify the kit. Our post on who sells farm technology in Kenya and what they deliver helps you work out which kind of seller you are dealing with.
Only one pass in the programme
The saving covers only part of the spray programme. In the Arkansas work, broadcasting residual herbicides gave better weed control than targeting them. Because of that, the researchers recommended broadcasting residuals and targeting only the post-emergence pass.
That caps the headline. The 43 to 59 percent applied to post-emergence herbicide, not the whole chemical bill. If a programme includes a broadcast residual pass, total chemical use will drop less than the post-emergence figure suggests.
Hardware brings another wrinkle. With a single-tank setup, growers must either apply residuals through the targeted system or make two separate passes. Two passes mean more fuel, more labour and more time in the field. A second tank or a second sprayer eats into the saving.
Before any quote, write down which passes your own programme has and which ones the camera would control. If that answer is one pass out of several, scale the expected saving to match.
Speed, field length and the nozzle
In the Arkansas system, the larger model can run at 15 mph. The more affordable model is capped at about 12 mph. Most of the small-plot research was done at 8 to 12 mph. The reason is simple: without extremely long fields, getting a sprayer to 15 mph is hard.
Kenyan field shapes matter here. Where blocks are short, the machine spends its time turning rather than at working speed, and the speed advantage on the spec sheet never appears. Ohio State's best management practices for boom spraying add that slower travel improves coverage. Higher speeds increase aerodynamic turbulence and boom instability, and so raise drift potential.
The camera also has to hand over to a nozzle that can behave. North Dakota State's guidance on drift-reducing nozzles says the smallest 10 percent of spray volume (the Dv0.1) should be 150 microns or greater. It also notes that ASABE standard S572.3 defines eight spray quality classes from extremely fine to ultra coarse.
That droplet logic is why nozzle-level control is possible at all. It is also why a switching nozzle must still produce coarse enough droplets when it fires.
Weather still decides whether the sprayer should run
A camera tells you where the weeds are. It does not tell you whether the air is fit for spraying. Our network's spray window score, computed hourly from station readings, gives a sense of how often conditions are poor. Across 7 stations, between 11 September and 11 October 2026, there were 5000 station-hours. Of those, 18 percent scored 60 or better on the 0 to 100 index, and 56 percent scored below 30.
That is a past-period record from our stations, not a statement about your block or a forecast. Still, it makes a point: if more than half of hours are poor, an expensive machine will sit idle waiting for the good ones. Its payback depends on hours it can actually work.
The Ohio State advice on common-sense spraying recommends flat-fan nozzles over cone nozzles for herbicides, because cone nozzles produce a much larger share of extremely small drift-prone droplets. It also names GPS, pulse width modulation nozzles and auto-guidance as advances worth taking advantage of.
If your question is when to spray, a weather station on the farm costs less than a camera. Our pesticide optimisation page covers that side.
What a Kenyan grower can do without cameras
Manual spot spraying is the cheaper route, and it gets priced from a rate table. The Kansas State weed control book carries per-acre rates and spot-spraying rates for brush control, with the costs attached, as the K-State brush control rates document shows. It is a pasture and brush publication. From it we take only the format: a spot rate sitting beside a broadcast rate, so the two can be compared on cost.
A Kenyan farm can build that same table for its own block. Scout the field, note where weeds cluster, spray those patches with the sprayer you already own, and record what you used against what a full pass would have taken. Your own records then become the evidence, the same job the Arkansas trial did for American growers.
We give no doses or application instructions here. Follow the label and what your agrovet or extension officer advises, and use only products registered by the Pest Control Products Board. For sprayers and how they are operated more generally, the FAO volume on equipment for pest and disease control is a foundation to work from.
Keep the weakness of this route in mind. It rests on whoever does the scouting, and it will not cover ground at the speed a camera does. On a small block, that trade is usually acceptable.
Advice is cheap to deliver, hardware is not
Kenya already has working rails for technical advice. PlantVillage's precision ag programme reached almost 500,000 people by phone and 9 million by TV, and sending 350,000 advisory messages a week costs just $1,800. In the Long Rains of 2022, 8.8 million SMS messages went to Kenyan farmers. Of 4,272 farmers surveyed, 3,577 said the forecast helped and 605 said it did not.
Forecasts and advice were what that programme was about, not spot spraying. We use it to make one point: a message telling a grower when not to spray costs a tiny fraction of a camera boom, and it lands with people who would never buy one.
So where does a camera earn its cost? On a large, long, mechanised block with a spray programme where the post-emergence pass is a big line, and where the owner will run sensitivity high enough to protect next season. Anywhere else, scouting, patch spraying and better timing come first. Nothing we hold supports a Kenyan saving figure, so before you sign, ask any seller for a trial on your own field, with weed counts the following season.
For the practical side, NuaSense has a longer guide. How to apply pesticides in Kenya: A safe spraying guide takes you through picking the right product and getting your spray timing right.