by Sally Colby
Farmers know their land. They’re aware of every field with below-average yields, where water accumulates and every spot that could use just a bit more lime. The challenge is making adjustments to only those areas without spending additional time or money.
Whether or not they use it, most farmers are familiar with the concept of precision ag – the use of information technology to aid in more precise management of the farm. Farmers who aren’t using this technology have hesitated because the size of their farm doesn’t warrant the equipment investment.
But if the price was right, would more farmers use precision ag? Ranveer Chandra believes they would. As principle researcher at Microsoft Research, Chandra is working on a project to help bring precision ag to farms of all sizes – without breaking the bank.
When the technique of site-specific, data-driven agriculture was introduced in the 1980s, it wasn’t widely adopted because data collection was expensive and time consuming. Not much has changed – Chandra recently saw precision ag equipment (sensors) at a cost of $8,000 for five sensors, an unreasonable price for most farmers to pay without knowing the return on investment (ROI).
Research on data-driven agriculture shows it can improve yields, reduce the cost of production and contribute to sustainability. Chandra’s goal is to see every farm in the world thoroughly mapped and overlaid with data such as the soil moisture level six inches deep, soil pH, soil fertility and numerous other data points. Detailed maps like this can be used to apply water, fertilizer and other inputs only when and where needed.
Chandra is leading a data-driven agriculture incubation project known as FarmBeats. He described FarmBeats as an end-to-end IoT (Internet of things) system – one that can take data from drones, cameras or other devices, send those data to the cloud and convert them to actionable insight. “The goal of the FarmBeats project is to bring down the cost of data-driven agriculture solutions from $8,000 to $80,” he said. “Based on our technology, I think we can get there.”
Chandra explained IoT by comparing it to the more familiar internet. “The internet was designed for human consumption,” he said. “You’re on a laptop using the internet, or on a phone using the internet. A human is doing something.”
IoT is the concept of connecting devices or objects to a synchronized network without human interaction. A few simple and familiar examples include devices used in homes such as robotic vacuum cleaners, automatic watering systems for lawns and gardens or lowering window blinds at a certain time of day. Chandra said the concept translates to agriculture and can help every farmer add precision ag to the farm. However, there are three challenges in making this technology reasonable and affordable: connectivity (lack of reliable internet throughout the farm), lack of direct power sources and limited resources.
“The first challenge is internet connectivity,” said Chandra. “The actual farm could be a few miles from the farmer’s home, and connectivity can be obstructed by crops, canopies and landforms.” Although most farms have internet access in the home, that access isn’t available throughout the farm – making it nearly impossible to transmit data.
Chandra has an answer to this problem: TV white space. Anyone who remembers switching from one television channel to another has seen the “fuzz” or empty channels between viable stations. That’s TV white space, and in 2010, the FCC made that spectrum legally available for use. Chandra said TV white space uses lower-frequency UHF signals that can cover uneven terrain and “break through” some obstacles. A white space device (router) allows WiFi access throughout the farm.
Chandra explained TV white noise channels are actually better than WiFi. “One of the key insights in the context of agriculture for this is that TV towers repeat a lot,” he said, adding there are more TV towers where there are more people. “The more empty channels, the more unused spectrum is available on the middle of the farm. We could be connecting drones, cameras, tractors – streaming a lot of data that you previously couldn’t get from the middle of the farm.”
The second challenge, a lack of a direct power source, can be solved in several ways. UAVs (drones) for such an application work automatically, cover large areas of land quickly and can collect visual data. However, UAVs have limited battery life (less than 30 minutes), are costly and come with regulatory concerns.
As an alternative to UAVs, Chandra experimented with tethered helium balloons that can go up 150’ to 200’. The balloons are equipped with smart phones and battery packs, and remain in the air for four to seven days to take pictures of a particular region. However, balloons are not stable, and the camera isn’t always facing downward. Chandra is working with new balloon options that might work for small farms.
The third challenge is limited resources. “If you want to build a very accurate soil moisture map, you need lots of sensors,” said Chandra. “You would need sensors every 10 meters, and that is expensive to deploy, manage and maintain.” Chandra said accurate maps can be built using linear interpolation and other statistical methods. A heat map of the farm can be created using a combination of weather data, drone video footage, sensors, cameras and handwritten notes. “You fly a drone or balloon, get the images and stitch together for maps,” he said. “Then take the sensor data, build a model and use that model to interpolate and come up with detailed precision map of the entire farm.”
Chandra said when asked to think of how they might use FarmBeats, farmers came up with numerous ideas including weed detection, livestock monitoring, crop yield prediction, yield variation and soil health analysis.
“The goal of FarmBeats is to allow farmers to view various aspects of the farm at any time and any portion of the farm,” said Chandra. “FarmBeats is an end-to-end platform for agriculture and includes sensors, UAVs, the connectivity piece (WiFi), along with the back end, which includes the cloud service and/or machine learning that provides insight farmers can use.”