FarmSense uses sensors and machine learning to protect crops from bugs – TechCrunch
Gnaw, Dig, Infect: Damage to agriculture from pests like the Japanese beetle (pictured above) exceeds $ 100 billion each year, according to the USDA’s Agricultural Research Service. And along with plant diseases, which exoskeletal buggers can also transmit, arthropods account for the 40% annual loss of agricultural production worldwide.
Enter FarmSense, an agricultural start-up based in Riverside, Calif. That is trying to solve the pest problem. The company is creating optical sensors and new classification systems based on machine learning algorithms to identify and track insects in real time. The key here: real-time information.
They claim that the real-time information provided by their sensors allows early detection and therefore rapid deployment of pest control tools, such as insecticides or biocontrols. Current mechanical traps used for monitoring may not provide important information until 10-14 days after the insects arrive.
“Some of these insects only live to adulthood for about five days, so by the time you know you have a problem, the problem has already taken hold and is now a more serious problem,” said Eamonn Keogh, co -Founder of FarmSense. “If you had known in real time, you could have located the procedure in one place and obtained a much better result, saving pesticides, saving manpower and preventing damage to the harvest.
How they can provide the essential information to achieve these best results is a bit complicated.
Currently being tested and studied in almond orchards in Southern California with a grant from Small Business Innovation Research, their new sensor, called FlightSensor, is best understood when you consider where Keogh got the idea from: James Bond and the espionage of the Cold War.
Keogh explained how Russian spies used lasers, placed on window panes, to pick up vibrations caused by people’s voices. Then a sensor would translate this information, providing rough information about what was going on in the room.
“With the same sort of cunning in mind, I imagined what would happen if an insect hovered over a laser… you would only hear the bug and nothing else.”
However, instead of reading vibrations, the FlightSensor uses light curtains and shadows in a small tunnel where insects are drawn to attractants. On one side of the sensor is a light source and on the other the optical sensor. The sensor measures the amount of light obscured, or rather the amount of light passing through it, when an insect flies inside. This data is transformed into audio and analyzed by machine learning algorithms in the cloud.
According to FarmSense, the sensor, designed to look like old analog devices for easy use by growers, does not pick up ambient noise, such as wind or precipitation.
“The quality of the signal is so beautifully clear and it is so muffled to the ambient sounds normally heard in the field,” Keogh said. “It’s basically a different way of hearing the bug, but when you put on headphones and listen to the audio clip from the sensor, it sounds like a mosquito or a bee flying. “
Keogh, professor of computer science and engineering at UC Riverside, specializes in data mining and works on new machine learning algorithms that FarmSense uses for identification purposes. Entomologists and field specialists, including co-founder Leslie Hickle, assist with development and deployment.
Shailendra Singh, the CEO of the company that has developed systems for wireless and cellular networks as well as security, works on the hardware side. He provided an operating price for each sensor, which will be billed seasonally, at $ 300.
The impact of this technology is clear. For farmers tending to fields large and small, real-time insect information would not only be important to their financial security, but would also allow them to conserve and potentially protect critical resources, such as soil health. .
But FarmSense claims to want to empower rural farmers who they say are disproportionately affected by insect damage.
Still, $ 300 per sensor per season is high, posing a potential risk for adoption and, therefore, for the technology’s ability to even tackle the insect damage problem in the first place.
One of the most difficult things for smallholder farmers is managing risk, said Michael Carter, director of the USDA-funded Feed the Future Innovation Lab for Markets, Risk and Resilience and Distinguished Professor of Agricultural and Resource Economics at UC Davis.
“Risk can keep people in poverty. This discourages investments in technologies that would increase income on average because the future is unknown, ”Carter said. “People with little wealth obviously don’t have a lot of savings, but they can’t risk the savings to invest in something that could improve their income and starve their families to death. “
However, he was optimistic that a technology like the FlightSensor could alleviate the fear of investments for small farmers, especially if the technology were paired with insurance to further protect them.
The technology also raises this question: is real-time identification really the best option for pest control? Speaking to research entomologist Andrew Lieb of the USDA Forest Service, that may not be the case. He explained that the main drivers of invasive insects – usually the most destructive to agriculture and forests – are travel and trade.
He expressed optimism for the technology as a way to control insect establishment, but ultimately believes the optimal strategy is to attack the problem even earlier. We should address the current import and export laws, how products are processed to eliminate pests, and maybe even pass travel bans.
Despite these concerns, there is no doubt that FarmSense’s technology is poised to make an impact. Even thinking beyond tackling financial insecurity for farmers and the threats to our global food chains, it could prove useful for tracking and disseminating essential information on disease-carrying insects, such as mosquitoes.
And with the continual disruption caused by COVID-19, it’s hard to imagine a world that is not fully aware of how the successes – or failures – of biosecurity impact our countless systems.
Looking at how invasions of non-native insects are expected to increase by 36% by 2050 and how population growth will put greater pressure on food production, innovative technology like the FlightSensor that improves our ability to understand and responding thoughtfully to threats is more than welcome.
As Carter said of every possible way agtech still benefits agriculture, “we have to be creative at these margins.”