Innovation & Technology Artificial Intelligence for Sustainability

These AI-Powered Drones Finally Bring Precision to Herbicide Application

Precision AI has developed the world's first AI drones for plant-level herbicide application — on a mission to create healthier, more sustainable, profitable farms.

By Scarlett Buckley | Published 1 year ago | About a 4 minute read

Image: Precision AI

As population growth, economic development and urbanization rise, our farmers are under continual pressure to produce more food. Currently, farmers are using intensive, industrial processes to keep up with demand, including the application of vast amounts of herbicides, pesticides, and fertilizers on crops. These chemicals pollute the world's soil, contaminate water supplies and threaten biodiversity.

With climate change also supercharging the growth of certain weeds that compete for nutrients, farmers must apply more herbicides to control them — which will further contribute to climate change; and thus, the problem will continue. Alongside the environmental issues comes the huge financial losses associated with the deployment of large-scale spraying; farmers must often spray 100 percent of their land to kill weeds that only persist on 3 percent of it. With pressure for farmers to sustainably produce enough food for a growing population, something’s got to give.

Enter: Canadian startup Precision AI — which has developed an autonomous, AI-powered agricultural drone that can effectively apply herbicides at plant level, at scale.

“As farms have evolved, they have consolidated — with farm sizes getting larger and larger; and that creates more complexity because now there are more acres to farm, more equipment to manage, and more decisions to make,” Precision AI’s VP of Business Development, Warren Bills, tells Sustainable Brands®. “On top of this, there is also the issue with accessing laborers — [with] urbanization, it's just harder to pull people into the rural areas where the work is.”

Drones such as Precision AI’s could provide farmers with the labor they need while providing real-time insights, making future predictions, and combining and collecting data to better understand a farm’s crops down to the individual plant. Overall, the technology could help farmers drastically reduce water use and costs, eliminate excess chemicals, and promote soil health.

“Farmers don't wake up every morning excited to spend millions of dollars on chemicals; but they have no choice because that's what is required to grow a healthy crop and be profitable. So, over time, this has created inefficiency; because they have to make large-scale decisions in a short period,” Bills says. “Our solution is very much targeted around the annual overspend of chemicals from broadcast spraying that we see occurring — real-time, edge-based, high-speed AI for precision allows us to reduce that overspend.”

Image credit: Precision AI

Precision AI recently won the Rising Star award for its use of edge computing — an emerging model that allows the drones to function without an internet connection or the cloud, mitigating problems with limited connectivity in the field; the drones have their own data center and process most of their data internally. This technology also enables the drones to provide near-instant weed identification. The drones can fly at 70 km/h and can detect and classify plants at an accuracy of 0.5 mm.

“When you think about the scale of this technology, taking a 20,000-acre farm and managing every individual plant, it's kind of crazy to think about — but that's what AI and computer vision can do,” Bills says.

Due to regional variations in crops and weeds, different AI models must be developed for different areas. Once all the necessary data for the crop has been received, Precision AI can build a model in close to six weeks.

“Once the model has got a standard data set in place and we’ve thrown some new data against it, it starts to learn,” Bills explains. “Then, it goes through a quality-control process; and we make some corrections, and it learns again. By about the fifth iteration, the model is really self-learning based on the new data that we give it from fields. Then, we have more quality-control procedures in place to see if we need to make any adjustments. We're in the startup space, so we continually have to question and make changes — but that’s because we're pushing the limits of what’s imaginable.”

After landing $20 million in equity and grant funding in 2021 and being selected for John Deere’s 2023 Startup Collaborator, Precision AI has gained commercial traction and aims to be operational by 2026. There are current legislation restrictions in Canada surrounding drone spraying; but this should change as governments become more open to newer agricultural technologies and after trials have taken place.

“In five years, I think our technology will allow us to have the best, most accurate data set in broadacre row crops in the world,” Bills exclaims. “We aim to be a leader in this space and expand into other types of applications — not just herbicides. We're keen to look at insecticide reduction, fungicide optimization, and biological use using drones. There's just so much more that this technology could be used for outside of herbicide.”