African drone company uses AI to give vital help to US fruit and nut farmers

JOHANNESBURG - South Africa’s Aerobotics is utilizing artificial intelligence (AI) in helping fruit and nut farmers improve crop yields. Although the Cape Town-based company only started nine years ago, it is already operating in 18 countries, with the U.S. being their largest market, followed by South Africa, Australia, Spain and Portugal. Its customers produce tens of millions of tons of fresh produce every year. 

California is now ground zero for Aerobotics – where the company has the biggest concentration of customers. On its 76,000 farms and ranches, sources agree, the state produces more than half of all fruit and vegetables grown in the U.S. 

The California Climate and Agriculture network recently warned, "Dependent on the weather and water availability, the state has much to lose if the worst impacts of climate change on agriculture are not avoided," the organization warned in a recent statement.

RESEARCHERS USE AI TO PREDICT CROPS IN AFRICA TO HELP ADDRESS FOOD CRISIS

Which is where Aerobotics has stepped in, using AI to reverse these trends by almost miraculously helping directly to increase not just the amount of produce grown, but also utilize the dwindling water resources more efficiently.

"Food security is a global challenge and everyone is being challenged to do more with less. Using the latest AI and different imagery sources, Aerobotics helps the fruit and nut industry make better decisions and improve yields," the company’s CEO James Paterson told Fox News Digital." 

He continued, "We work with a range of fruit and nut producers across the U.S., from citrus and table grape growers in California, to apple producers in Washington, to nut growers in Arizona and New Mexico. We have mapped over 600,000 acres of farmland in the U.S., and growers are uploading over 1 million images of fruit per month through our platform, using our system to scale their knowledge."

WHAT IS ARTIFICIAL INTELLIGENCE (AI)?

Paterson, who operates from offices in Cape Town and California, grew up on a fruit farm in South Africa, witnessing firsthand the harsh risks involved in fruit production. This planted the proverbial seed to find a way to use data to improve operations and knowledge.

He worked on cutting-edge AI and drone technologies when pursuing a master's degree at the Massachusetts Institute of Technology, learning how to address agricultural challenges, and then teamed up with Benji Meltzer, an expert in computer vision and software systems, to found the now 60-strong Aerobotics team.

Drones and mobile phones running AI software are operated by farmers and professional drone pilots to yield data about both fruit and trees. 

The porch or balcony favored by farmers worldwide in South Africa is known as a "stoep." This, Paterson proudly proclaims, is "farming from a stoep," as, when using drones, the farmer can evaluate his produce from his armchair.

In this case, AI can perhaps be accused of thinking, as the Aerobotics software and AWS, or Amazon Web Services, servers it feeds data and images to use this information to report on the health and status of fruit, and predicts crop yield. The information received helps planning for packhouses, sales teams and retailers. 

And it saves huge amounts of time, lopping hours off chores such as checking out pesky pests: AWS claimed in a statement that the AI system has cut down monitoring every tree for pests and diseases on a 50-hectare farm from an entire day to just 20 minutes. 

BIPARTISAN LAWMAKERS EYE AI SAFEGUARDS FOR US AGRICULTURE INDUSTRY

Imagery is "analyzed by AI models to detect individual fruits, and calculate various metrics including size, color and external quality or blemishes," Paterson told Fox News Digital.

"This data undergoes analysis through hyper-localized forecasting models to project the data forward to harvest."

"As data accumulates on a farm, the models are fine-tuned to that specific environment. Essentially, the AI models learn and adapt to localized growing conditions, enhancing forecasting accuracy and enabling comparisons to previous years," Paterson added.

Another AI program produces a digital model of each tree on the farm, at scale, tracking it over time. "Each tree is conceptualized as a factory that can be optimized to produce the highest quality fruit. Data is gathered by drones equipped with thermal and multispectral cameras, operated either by the grower or through our third-party pilot network," added Paterson.

In this increasingly water-scarce world, the Aerobotics AI also detects irrigation issues such as leaks, pressure problems and blocked water lines or pipes. The software assists with fertilizer usage and replanting damaged or missing trees. Typically, when farmers lose a tree, perhaps through disease, they have to wait five or six years for a new tree to start fully producing, but with this AI, early prediction is possible, ensuring farmers get back into production within a year.

U.S. food security is also improved as the AI utilizes per-tree data to determine crop insurance policies and safeguard growers’ production.

This data-driven approach helps in the production of high-value fruit, including citrus, apples, grapes, cherries, kiwis, table grapes and pomegranates, and nuts such as almonds, pecans, and pistachios.

"We've started using drone imaging to monitor tree health in our orchards," Aerobotics customer Matt Allred of Arizona’s North Bowie Farming, a pecan nut producer, told Fox News Digital. "As a result of looking at the drone images, we could see which blocks had lower health ratings and apply treatments over time."

"Multiple flights over time show these blocks' health catching up to the control blocks after intervention. The drone flights help us measure this across hundreds of acres, not just one small block. Seeing the improvement in the health uniformity of our blocks is what really sold me on the technology."

"AI plays a pivotal role in our business and to our customers," Aerobotics’ Paterson concluded. "AI enables us to construct models that generalize, learn, and operate effectively at scale. Using AI and imagery, we are able to increase efficiency of data collection by more than 10 times."