Agriculture contributes a third of the world’s greenhouse gas emissions as it strives to feed a hungry world. Precision farming and regenerative practices, guided by artificial intelligence, can help us avert a global food crisis and mitigate climate change. Spearheading this effort are women in key roles.
By Seabright McCabe, SWE Contributor
“It’s hard to comprehend that in the next 50 years, we’ll have to produce as much food as we have ever produced in human history,” Megan Clark, Ph.D., former chief executive of the Australian Commonwealth Scientific and Industrial Research Organization, said in 2009.
That alarm has only grown louder. The world’s population will grow to 9.7 billion people by 2050, according to the United Nations, and global food demand will increase by up to 70%. And the world isn’t gaining more arable land; it’s losing it — to soil erosion from unsustainable agricultural practices, pollution, and urbanization.
Transforming food systems could cut global emissions by about a fifth of what’s needed to avoid the worst effects of climate change by 2050, according to the Global Alliance for the Future of Food. Yet strategies for doing that are “startlingly absent” from most countries’ plans for tackling climate change.
“We are facing unprecedented challenges in the agrifood value chain,” Claudia Rössler, agriculture – strategic partnerships, principal program manager, Azure Global Engineering, Microsoft, said. “While crop yields grew significantly over the past 100 years using modern inputs, advanced seed breeding, and biotechnology, they are now reaching their limits. Technology bears the promise of fueling the next agricultural revolution.”
You can’t manage what you can’t measure
“It’s all about decisions, and good decisions are something this industry really needs,” Rössler said. “With about 600 million smallholder farmers producing 80% of the world’s food, 2 billion people are dependent on the efficiency of the system. AI will change how we make decisions on the farm in the future, enabling farmers to monitor data that predicts how things will happen before they happen.”
Artificial intelligence excels at performing simple tasks more accurately, and at speeds no human can match. “What’s really exciting is that it can overlay tons and tons of data,” Rössler said. “AI helps us understand the complex correlation between weather, soil, crops, and crop management. These data-driven techniques with AI can simultaneously boost agricultural yields, lower carbon footprint, and save costs by reducing the use of water, fertilizer, and pesticides.”
One of the biggest uncertainties in farming is weather, which determines so much of a farm’s success or failure. “We want to help farmers prepare for its unpredictability,” Rössler said. “With the power of cloud, the Internet of Things (IoT), and AI capabilities, farmers can receive microclimate predictions.” AI can also help farmers learn the soil conditions at different depths in precise locations, and predict yield or livestock conditions, and even help foresee potential risks such as crop diseases or pests.
For farmers facing risk from many directions, AI is a natural fit. “What we’re seeing today is more intelligent equipment and more connected farms overcoming the challenges of the rough environment out there,” Rössler said. “And we haven’t really scratched the surface of what’s possible.”
AI for rural food security
In the developing world, the poorest farmers work the land with the fewest resources, and the least technology. Satellite images using weather data can help increase efficiency, but a little more AI could go a very long way. “It always inspires me when I think about how much can be done with technology for rural farming,” Rössler said, describing a number of AI-driven initiatives addressing food security and limited resources.
“A couple of years ago, we partnered with the International Crops Research Institute for [the] Semi-Arid Tropics (ICRISAT) in India to deliver planting advice in the form of text messages to rural farmers,” she said. “With as little as 10 messages throughout the planting cycle, starting from the best planting date, farmers were able to improve yield by 20-30%. And for those who can’t read, AI takes that information, translates it into the local language, and calls farmers with advice so they can make better decisions.”
Another Microsoft partner, SunCulture, develops irrigation and farming technology solutions to help smallholder farmers in Africa increase crop yields and earnings. Its solar-powered irrigation systems save hours that would otherwise be spent irrigating fields by hand. Additionally, SunCulture’s AgOptimized app compares sensor data about current conditions with historical climate models, then provides detailed forecasts and recommendations for individual farm plots.
And Microsoft 4Afrika is partnering with the International Finance Corporation to build KuzaBot, a mobile chat platform that speeds information about sustainable agricultural practices into the hands of farmers.
Many more projects are being developed through Microsoft’s initiative, AI for Earth, which provides grants to researchers, universities, startups, and individuals to build AI models for sustainable farming. “It’s tremendous, the innovation coming from startups,” Rössler said.
While each project differs in nature, they all have the same goal: to provide a better income and livelihood to as many of the world’s nearly 600 million smallholder farms as possible. “Global food security will improve significantly if we can help them not only feed their own families, but their communities and beyond,” Rössler said. “We need to bring technology to the farm and enable those who need it most.”
Irrigation goes digital
Agriculture accounts for nearly 70% of the world’s water consumption, and knowing exactly where and how much water is needed conserves that precious resource. With that in mind, Schneider Electric, New Zealand-based WaterForce, and Microsoft Azure developed a cloud- and IoT-based, smart irrigation solution called SCADAFarm, which allows farmers to move irrigation pivots nimbly, efficiently, and sustainably, all from a desktop or mobile device.
“It’s a good example of how you use external data to gain insight,” Rössler explained. “One simple layer shows weather data over the farms. Then you can drill down to an alert from an irrigation system warning of potential damage from forecasted high winds, telling farmers when they should align irrigation equipment with wind direction to avoid damage. Go a level deeper, and it monitors reservoir level, river height, and electricity prices, so it can advise the best time for filling reservoirs.”
Smart irrigation pivot systems also monitor soil moisture and plant root saturation 24/7, telling growers how much water is needed and where, and tracking the data so the system continuously learns.
Data-driven irrigation decisions pay off in results. Using SCADAFarm, the farmer was able to reduce water use by 30% and electricity use by 50%.
Autonomous greenhouses
Controlled environment agriculture is another area where AI can help growers optimize light, temperature, humidity, and carbon dioxide concentration for healthier, more productive plants.
In 2018, Microsoft Azure worked with Wageningen University & Research (WUR) in the Netherlands to equip a greenhouse with soil sensors and cameras, producing the world’s first cucumbers grown completely with AI. Teams competed with expert growers in the university’s annual Autonomous Greenhouse Challenge, to see who could get the biggest harvest.
“Our team outperformed the others in productivity, and most of the teams with sustainability factors like heat and water use,” Rössler said, pointing again to AI’s predictive ability. “For example, people tend to vent a greenhouse only when it gets too hot, and don’t proactively address challenges right away. AI can tell them ‘it’s going to get too hot — you have to do something right now.’”
AI-driven, indoor crop production can not only save resources and outperform human growers, but it can also reduce hands-on labor during labor shortages. The challenge, now in its third year, is to grow a crop of lettuce completely by remote control. WUR calls it “an important step toward computer-operated, indoor farming, to better feed the world with healthy products while making optimum use of energy and water.”
Digging Deep for Soil (and Human) Health
Maryam Rasoulidanesh, Ph.D., machine learning specialist at Vancouver-based Terramera, grew up with an interest in STEM and agriculture, building systems to monitor her houseplants and creating algorithms to track their growth and health. “I wanted to have a positive impact on the world, so engineering was an obvious choice, and Terramera was a great option for me,” she said.
Terramera, whose name means “our earth,” is bringing clean tech to agriculture, by combining computational chemistry, sensor, satellite, and farm activity with AI. The company’s mission is to drive transfer learning, economically scalable accuracy, and more sophisticated deep learning models in order to transform Canada’s agriculture industry.
Seeing into soil from the sky
Unsustainable practices, such as overtilling of fields, growing monoculture crops, and overuse of chemicals, depletes soil and causes erosion far faster than new topsoil can form. With the race on to save arable land, part of Dr. Rasoulidanesh’s work is collecting data on soil health with high-tech drones.
How can a drone “see” into the soil? “Our drones use novel sensor and imaging technology,” she said. “As humans, we only see a small range of wavelengths with our eyes. But there’s a much bigger range beyond that, so imaging systems such as hyperspectral, multispectral, thermal, lidar, and high-resolution RGB [red green blue] can show us the full picture when it comes to soil health.”
Soil’s reflectance properties are determined by moisture and iron oxide content, texture, and structure. “We find different reflections and impacts of soil components,” Dr. Rasoulidanesh said. “Using machine learning models, we bring that data to life and show what’s actually being seen by these sensors.
“Drones gather accurate qualifying measurements that show us what’s going on under the soil surface that farmers can use to better sustain their plants,” she continued. “For example, hyperspectral imaging lets us accurately measure soil carbon content. Drones also can measure soil health at scale, covering vast areas quickly. High-resolution mapping of soil components can be used to improve precision environmental management of farmlands.”
Drones have huge potential for precision farming. They can monitor anything from livestock to plant health. They’re effective for assessing damage after a storm or natural disaster. They can identify areas in a field that need replanting, thinning, or pruning and can even find and spot spray-invasive weeds. All of these advantages save time, energy, and resources for farmers operating on razor-thin margins — time that could be better spent on other tasks.
An experienced drone pilot, Dr. Rasoulidanesh enjoys the days spent flying in the field. “Especially in fall and spring, when there’s not much vegetation so we can see the soil itself,” she said. “It’s very different from my normal job, sitting at a computer and working with my codes. The missions are a total, harmonious collaboration, with everyone on the team doing their part.”
Dr. Rasoulidanesh strongly believes in regenerative, sustainable agriculture. She sees her role as a “link in a generational chain” of protecting the environment as a whole. “Our parents and grandparents worked toward the same goal of awareness that’s happening right now,” she said. “I hope that when our kids talk about our generation, they know we cared about the environment and were the ones that saved the earth for them.”
Green chemistry and soil/plant health
Conventional “spray and pray” methods, which blanket crops with synthetic weed- and pest-controlling chemicals and fertilizers, are another unsustainable practice — 50-90% of it isn’t even taken up by plants, but runs off into soil and water, disrupting the nutrient balance of delicate ecosystems and soils, and impacting human health. Green chemistry, an area of chemistry and chemical engineering that minimizes or eliminates the use of hazardous chemicals, is the antidote.
“We are what we eat,” Forum Bhanshali, research scientist and in vitro biology team lead at Terramera, said. “Chemicals cause so much damage in our food, and contribute to all sorts of illnesses, from cancer to neurodegenerative problems. Replacing chemicals with bioactives can do so much in the fight against these diseases.”
Bhanshali leads a high throughput, in vitro biology team using an R&D platform to unlock the next generation of formulation development of effective, environmentally friendly agricultural input products. Part of her work deploys Terramera’s proprietary, green chemistry technology, Actigate, which speeds selection of candidate formulations.
Automation helps accelerate the stages of product development. “We test the groundbreaking formulations our chemistry team produces,” she said. “The high throughput screening is powered by automation, robotics, machine learning, and computational chemistry modeling. All of these tools help us quickly test thousands of formulations for their target efficacy.”
The formulations her team generates aim to reduce the use of pesticides by eliminating synthetic chemicals and replacing them with organic bioactives. Bhanshali is currently working on solutions for dealing with insects and plant diseases, screening their effectiveness against specific target pests.
“It’s like a feedback loop,” she said. “My team generates data from high throughput workflows. Then the next team applies machine learning and computational modeling tools, which then further lead to even more effective screening and selection of the best candidates, through deep learning of the data,” she explained. “What I do supports the ‘in silico’ or computer simulation platform, so that we can be even more focused and more confident in our efforts.”
Growing up in India, Bhanshali learned about DNA’s double helix in 10th grade biology class, and Watson and Crick’s Nobel Prize. “There was a paragraph about Dr. Rosalind Franklin, and just one line about her research being essential to DNA’s discovery, but saying she didn’t also receive the Nobel because she died at an early age,” she said. “That didn’t seem right to me, so I researched more on my own — and found that despite the discrimination she faced, she relentlessly pursued her curiosity. After that, I started saying ‘I want to be a scientist.’”
That inspiration set her on the path to four degrees in biotechnology, microbiology, and pathogen cell biology, and to her work at Terramera. “If we can reduce the use of synthetic chemicals,” she said, “we can really change the world, and transform the food we know today.”
Connecting farms to the cloud
Rössler is proud of the work being done with FarmBeats, a platform that creates analytics models for precise and sustainable farming. FarmBeats aims to simplify collection and aggregation of weather, satellite, OEM, and sensor data, using low-cost sensors, drones, and vision and machine learning algorithms.
Dancing Crow Farm is one of FarmBeat’s many success stories, and was its pilot project. The goal was to produce more precise mapping and better recommendations to help the farmer better control soil pH and reduce water use.
As is common in rural areas, the farm had no internet connectivity. But engineers were able to transmit field data from sensors and cameras placed at intervals to solar-powered routers, using television “white space” (the empty bandwidth between TV channels) to create an IoT connection to the farm owner’s computer. Enabled by sensor-level data and better predictions, the farm ended up using 30% less water and 44% less lime for soil pH. FarmBeats also tracks moisture levels that determine when to plant, and can find areas with inadequate drainage, which impacts produce quality.
What the future holds
Looking down the road, Rössler is optimistic about virtual reality’s potential to benefit agriculture. “Over the course of the next 20 years, we will see the rise of preemptive analytics,” she said. “Imagine creating a virtual model of a farm, plant, or animal that allows you to simulate different conditions to make informed decisions, like determining what to plant where to get the best yield or the best return on investment.”
While her role gives her a 360-degree view of the agriculture industry, Rössler also finds inspiration in the farmers themselves. “I’m impressed by how incredibly open to technology many farmers are — both young and old,” she said. “They are risk-takers by nature, and one of my most enlightening experiences was with a 92-year-old farmer who approached me with questions after a keynote. And a young farmer we work with who proudly showed me how he stays on top of conditions on his wheat farm over his mobile phone.”
An ambassador for Women in Food and Agriculture, Rössler is also committed to a future with more women in the industry. “Women have traditionally been left out when it comes to passing down generational knowledge,” she said. “I see in them a higher willingness to use technology and practice new, more sustainable farming, to network and crowdsource innovation, and use a more collaborative approach to achieve better and sustainable yield results.”
Drones instead of crop-dusters; sensors instead of spades; clouds of data bigger than clouds pouring rain — is a world where agriculture is managed by AI technology on the horizon? Like farming itself, time, climate, and the tools we are building will tell.