Microsoft opens sources for its ‘farm of the future’ toolkit

FARMINGTON, Wash. – The gentle hills here in eastern Washington have long produced bumper crops of wheat, barley and lentils.

Fifth-generation farmer Andrew Nelson is adding a new bumper crop to that bounty: data.

It collects it from sensors on the ground, drones in the sky, and satellites in space. They give Nelson information about his farm at different points, every day, throughout the year: variations in temperature, soil moisture and nutrient levels, plant health, and more.

Nelson, in turn, enters that data into FarmVibes Project, a new set of farm-focused technologies from Microsoft Research. Starting today, Microsoft will open source code for these tools so that researchers and data scientists—and the queer farmer like Nelson, who is also a software engineer—can leverage them to turn farm data into actions that can help grow yields and reduce costs.

The first open source version is FarmVibes.AI. It is a sample set of algorithms intended to inspire the data science and research community to advance data-driven agriculture. Nelson is using this AI-powered toolset to help guide decisions at every phase of farming, from before the seeds hit the ground to long after the harvest.

FarmVibes.AI algorithms, running on Microsoft Azure, predict the ideal amounts of fertilizer and herbicide Nelson should use and where to apply them; forecast temperatures and wind speeds in your fields, reporting when and where you plant and spray; determine the ideal depth to plant seeds based on soil moisture; and tell him how different crops and practices can keep carbon sequestered in your soil.

“Project FarmVibes allows us to build the farm of the future,” said Nelson, who partnered with Microsoft Research to turn his 7,500 acres into a proving ground for Project FarmVibes. “We are showing the impact that technology and AI can have on agriculture. For me, Project FarmVibes is saving a lot of time, it’s saving a lot of cost, and it’s helping us control any problems we have on the farm.”

The new tools grew out of Microsoft’s work with big clients like land of lakes Y Bayern to integrate and analyze data. Project FarmVibes reflects the latest research in sustainable and precision farming.

By opening up its latest research tools, Microsoft wants to extend them well beyond Washington to help tackle the world’s urgent food problem, said Ranveer Chandra, managing director of Research for Industry.

By 2050, we will need to roughly double global food production to feed the planet, Chandra said. But as climate change accelerates, water levels drop and arable land disappears, doing it sustainably will be a big challenge.

“We believe that one of the most promising approaches to address this problem is data-driven agriculture,” he said.

At Microsoft, we’re working to empower farmers with data and artificial intelligence to increase their agricultural knowledge and help them grow nutritious food sustainably.

Research pays off

Until recently, Nelson’s farm was like many others around the world. He had internet at his house, but the Wi-Fi signal ended outside his door. His 7,500 acres were a dead zone.

You are now using a solution from Project FarmVibes, called FarmVibes.Connect, which will eventually be open sourced by Microsoft to bring connectivity to remote and rural locations. It provides broadband access through TV white space, the unused spectrum that flickers like “snow” between channels. Today, Nelson has a solar-powered TV-blank antenna that acts like a Wi-Fi router, but can cover most of his farm.

That connectivity has allowed it to pull information from the FarmVibes.AI suite. Now available on GitHubFarmVibes.AI includes:

• • Async Fusion, which combines images from drones and satellites with data from ground sensors to provide insights. For example, Nelson uses Async Fusion to create nutrient heatmaps from multispectral drone imagery and soil sensor data. These maps are used to vary the rate at which you plant seeds and apply fertilizer, which can increase yield and prevent overfertilization. Async Fusion can also create soil moisture maps from sensor data on Nelson’s farm. These maps tell Nelson how deep to plant his seeds and in what order to plant his fields. As an added benefit, they can help keep tractors and sprayers from getting bogged down in dirt.
  • space eye, which uses AI to remove clouds from satellite imagery. This helps Nelson fill in the gaps in areas he hasn’t explored with a drone. He can then feed these images into AI models that can identify weeds, helping him create maps to distribute herbicides only to areas that need them. And even when he sprays, these maps let him vary the application rate, delivering more volume to heavily weeded patches and a lighter load elsewhere.
  • deepMC, which uses sensor data and weather station forecasts to predict temperatures and wind speeds for your farm’s microclimate. In the Nelson area, the local weather forecast predicts what conditions will be 10 meters above the ground. “Well, I don’t care what’s 10 meters above the ground,” she said. “I care what is where my crops are.” Earlier this spring, Nelson was preparing to spray his wheat fields. He checked the forecasts for the proper weather window; the plants would be damaged by the herbicide if you sprayed them in sub-zero temperatures. The local forecast looked promising, but DeepMC predicted a freeze. Stopped spraying and woke up to frost.
  • A what-if analysis tool that estimates how various farming practices would affect the amount of carbon sequestered in the soil. Today, Nelson uses these what-if scenarios to improve the health of his soil and increase yields. But he plans to use them to enter carbon markets, which pay farmers for practices that keep carbon dioxide locked up in the soil instead of entering the atmosphere.