Dr. Ranveer Chandra – Data-Driven Farming: Taking the guesswork out of agriculture
Dr. Ranveer Chandra, chief scientist at Microsoft Azure Global, joins us to explain the company's FarmBeats project and how it is taking the guesswork out of agriculture. According to Chandra, data-driven agriculture makes operations more profitable and sustainable — but some farmers rely simply on their instincts and personal experience to make decisions about their crops or animals. Chandra and the FarmBeats project seek to provide these farmers with data and insights about their operations that will help them make informed decisions for their production.
The following is an edited transcript of the Ag Future podcast episode with Dr. Ranveer Chandra hosted by Tom Martin. Click below to hear the full audio.
Tom: Dr. Ranveer Chandra is the chief scientist at Microsoft Azure Global, and among the projects he is currently leading is FarmBeats, where he is principal researcher.
The goal is to enable data-driven farming with the aim of improving yield and reducing cost while ensuring sustainability.
Dr. Chandra joins us from Seattle. Thanks for being here, Dr. Chandra.
Dr. Chandra: Thank you for having me.
Tom: And let’s begin, if we could, by laying out some fundamentals. There is this imperative to increase the world’s food production by 70% by 2050. What is driving this effort?
Dr. Chandra: As you know, the world’s population is increasing, there is more urbanization, more and more people are moving to cities, and there is a need to feed this growing population of the world. That is, the amount of land is not increasing — the amount of good land — the water level is, like, going down, the soil is not getting any richer.
So then, the question is: How can we feed this growing population of the world to — and not just feed them food, just give them food, but give them good food and grow this food in a sustainable way?
Tom: I mentioned data-driven farming. How does that differ from traditional agriculture?
Dr. Chandra: Yeah. A lot of decisions a grower makes are based on guesswork. So, when I started this project, I went and talked to several growers. What I realized is that growers, these producers, they know a lot about their farm; they’ve been farming there for decades and, in some cases, even centuries. Like, there was one farmer who could feel the soil and say what’s going on. There was another farmer who could taste the soil and say what’s going on. But even though they knew a lot about their farm, a lot of decisions they made were based on guesswork — like, for example, how much water to apply, where to apply, what seeds to put, when to put out, when to harvest, when to plant. A lot of these decisions are based on guesswork.
If this guesswork was replaced with data and data-driven insights like a vision is, if you could take a farmer’s knowledge and augment it with data and data-driven insights, you could enable a future of agriculture that is more productive. You could grow more that is more profitable because you would reduce cost; you would use less inputs, less water, less pesticides. It’s also better for the environment for the same reasons: that you’re not wasting water, you’re not wasting pesticide.
So, that is what I mean by data-driven agriculture: the ability to use data to augment the farmer’s knowledge, so that farmers are more profitable and they’re using sustainable practices for their farming.
Tom: I’m imagining a field of crops, and perhaps one corner of it, or one section of it, is usually drier than other sections of it, and that can be difficult to detect. Is what we’re talking about here being able to detect areas that need more moisture or less moisture, for example, that kind of thing?
Dr. Chandra: Yes. So, you’re then able to, like — for example, meaning which, I believe, if you would farm properly, you will have sensors everywhere. You would be able to see what’s going on throughout the farm. Wherever there is stress, you’d be able to respond to it in a timely way. And that’s very difficult to do. Like right now, mostly, you don’t know what your entire farm looks like. With data-driven agriculture, we can give you that view of what your farm looks like at any instant in time. Not just in terms of imagery; maybe even in terms of sensors. For example, what might be happening below the surface.
So, one, you’ll be able to detect what’s going on, and second, you might be able to diagnose why are you seeing what you’re seeing, so that you can then take corrective action and fix that — and exactly to your point that there could parts of the farm that are very different than the others in different ways. For example, it could be in terms of crop stress, it could be in terms of nutrients, it could be in terms of moisture, water stress. And to be able to flag that in a timely way so that you can take corrective action is where data can help.
Tom: Okay. Tell us about the FarmBeats project. How does this work?
Dr. Chandra: The FarmBeats project started in research, and the goal was exactly what we discussed: How do you take guesswork out of agriculture? How do you enable growers to take more decisions based on data and data-driven insights?
And toward that, we have been developing various methods, both in terms of the Internet of Things, being able to get data from the farm, from different sources about the farm. For example, you can have sensors and drones and tractors in the farm, you could have satellite data, weather data. All of these data being able to bring all of that together in one place in the cloud, and bring the benefits of the cloud in artificial intelligence on top of that data, so that you can then start driving insights to the growers.
And that is what we have been doing in the FarmBeats project: coming up with new ways to bring data from remote parts. For example, using new connectivity technologies, being able to leverage a news TV channel to send and receive data, so that even if you have no internet right now, you could then use this new method to start sending data using technology such as edge compute.
That is, if you have a camera somewhere in the farm, sending all that data to the cloud will take a long time and will need a lot of bandwidth. You could, instead, be doing a lot of processing on the farm itself using edge compute. And then, when you bring all of these data to the cloud, you’re bringing new artificial intelligence tools to be able to merge different data streams.
For example, you could be having very few sensors on the farm. We then use artificial intelligence to start predicting what the sensor values would be in other places where you don’t have sensors so that, at low cost, you can then start building these detailed maps of the farm.
So, with the FarmBeats project, that’s what we’re building. We are building this platform for data-driven agriculture, the ability to bring data from different data variety of streams in a way so that you can then start running AI, artificial intelligence techniques on top of that data to drive new insights, to be able to predict things that you otherwise wouldn’t know.
Tom: I wonder, how can you gather data from farms that have no access to power in the field — or connection to the internet, for that matter?
Dr. Chandra: Yeah, and that’s what makes agriculture so fascinating. As you know, my background is in computer science. I’m a computer scientist, and I work in different areas, not just agriculture.
When I started working in agriculture, I realized that the agriculture poses very interesting problems to technology. For example, these farms, many of them do not have any internet; many of them don’t have power. That is, you don’t have power outlets so you could plug in your devices. Well, people say you could use solar panels, but then you have to spend the winter in Seattle to realize that it doesn’t work. We get very cloudy winters.
So then, to address that problem, we’ve been developing new methods. For example, one way to get internet from the middle of the farm is using the technology I mentioned called TV white spaces. What the TV white spaces enable is — imagine a Wi-Fi that can go several miles, and one of the ways you could get that is if you take a white signal and put them in noisy TV channels. These are TV channels you watch using antennas, over-the-air antennas.
You know, when you browse through TV, on certain channels, you get a transmission, and on other channels, all you see is white noise. The interesting thing about that is that most of these TV towers are in the cities. So, if you turn on a TV in the middle of a farm, most of the channels are just white noise. While that’s not great news for a grower who wants to watch TV, it’s great for someone who wants to use that unused TV channel for sending and receiving data. So, even if you have 20 TV channels that are available, we are talking of over a few hundred megabits per second of available capacity in the farm.
So, this is one way in which we are bringing connectivity to the farm. The other thing we are doing is bringing edge compute. That is, this is a scenario where, if you have barns where there are multiple cows and you have cameras, rather than sending all the camera data to the cloud, you could have edge compute. Imagine a small computer sitting in the barn or in the farmer’s house that takes all of this data, the camera data, and runs the processing over there itself, so that you could be sitting in your house and monitoring how your cows are doing. You could be getting alerts if a cow is sick, if the cow is not moving very well.
And these are, again, things that could be enabled because of these new technologies: TV white spaces, edge compute, the Internet of Things. And we’re bringing all of that to agriculture.
Tom: This is fascinating. Have these technologies been deployed?
Dr. Chandra: So, these technologies, some of them are Microsoft products. Like, for example, we have Azure IoT, Azure Edge, the Azure Stack Edge. And in the context of agriculture, we are working on some of these technologies. We have farms where we’ve deployed TV white spaces, we’ve deployed edge compute, we are doing all of this intelligence on top. And there are various farms in the U.S. and abroad where we’ve deployed this and shown the feasibility of this technology.
Tom: How are you using drones and sensors to map such key data as soil moisture and pH levels?
Dr. Chandra: One of the things we want to know that the growers want to know is: How do certain soil properties (that) are in the farm produce certain weather properties (that) are in the farm? And you could get that information using sensors. So, these are sensors, for example, you could get from our partners, such as Davis Instruments special instruments.
But the question, then, is where do you put these sensors? And so, one of the algorithms we have, an artificial intelligence algorithm we have, is it will tell you — once you give it the farm boundary, we then get the satellite image for that farm. And then, for that farm historically, we look at the satellite imagery for that plot of land and then estimate the best places that you need to put sensors.
You could — say you have, say, three sensors. You can then use artificial intelligence to decide the best places you put those sensors. Once you put those sensors in the farm, the data then starts going all the way to the cloud. What that means is you could be anywhere in the world and you could turn on your phone and you’ll be able to see what your farm looks like, what those sensor values look like.
But the question, then, is if you have, say, a thousand acres of land and you put just three sensors, it will just tell you three points in the farm. You want to know more; you just don’t want to know those three points where you intelligently place the sensors.
This is where we use, again, artificial intelligence, where we combine the sensors with aerial imagery, say, from drones or satellite. The way we use it is, our key insight — and you would be able to relate to it — is that if two parts of the farm look similar, they are likely to have similar values. When I say look similar, it’s not just in red, green and blue, but in multispectral or high-spectral imagery, they are likely to have similar values.
And we incorporate this intuition, this insight, in an artificial intelligence model again, where we combine the small, the sensor values with aerial imagery, say, from satellites or drones to build a heat map of what your farm looks like. So, with very few sensors, by bringing the latest in technology, you’re able to visualize what the problems are in different parts of the farm.
And then, partners, the companies that we work with, could then start building solutions on top. Once you have this map, you can see how you could build several agricultural advisory solutions — for example, for irrigation or fertilizer management or others — on top of this framework to start providing insights to the growers.
Tom: We’ve been talking about farming that revolves around crops, around plants. How is the technology being used in animal husbandry — poultry, for example?
Dr. Chandra: Yeah, that’s a very interesting question. And I recently gave that talk at the Alltech poultry conference. You could — the same ideas that I talked about could be used for poultry as well.
For example, I talked about how, in a barn where you have cows, you could be using cameras to determine how the cows are moving around. Imagine, like, a baby monitor, but for cows. You could then get notification when a cow is sick so that you can then provide timely intervention to manage the cows.
The same things could be used for poultry as well. That is, a lot of times, when you’re in chicken coops, you might not, like — especially when the chickens are young — you want to know how they’re doing, you want to know that the conditions are right. You might want to put sensors in the farm; you might want to put temperature sensors (or), in some cases, humidity sensors. So, (something) similar to kind of an IoT system could be beneficial for poultry, too.
In fact, we are taking a step further. One of the projects we did, this was with the University of Washington, is where we’re looking at acoustic information that — instead of cameras, where cameras can’t see through obstructions, we were looking at putting microphones in chicken coops. And the system that the students spearheaded was called [clucky?] eye, where they had these microphones, and then, by looking, by observing the sound patterns of chicken, you could tell when a chicken is in stress. And that would be information that you could flag to a farmer (and) say, “Hey, go, there’s something wrong. Maybe there is predation going on; the chickens are not happy, for whatever reason.” And this is, this could help poultry farmers be more profitable (and) avoid damage which, otherwise, which will be just hard to monitor.
And there are many more new cases, too, and this is where I really enjoy talking to farmers. So, if there’s any farmer who’s listening to this podcast and would like to discuss ideas of how technology could be used for poultry farming or other farming practices, I would love to get them a comment (or) chat.
Tom: How they can reach you?
Dr. Chandra: Yes. The best would be to either add me on LinkedIn or send me an email on ranveer@microsoft.com, and I try to get back on email. So, I would love to talk to more growers, more people in the agriculture community, to discuss ways in which we can bring the technologies I talked about and many new things to help farmers be more profitable and practice the most sustainable practices.
Tom: Talking about profitability, what about the cost of purchasing and deploying these technologies. How do you make it affordable to small farms, in particular, that might be interested but may be discouraged by the cost?
Dr. Chandra: This is where we want — well, that’s one of the other problems with agriculture, is you want to be providing a lot of these insights, and you want to get data from a lot of places that do not have great connectivity, but you also want to bring the cost of these devices down to a point where they’re affordable. And that’s been one thread of what we have been trying to pursue as part of the research, is (what) we (can) use to bring down the cost of these data-driven agriculture technologies.
For example, I talked about leveraging TV white spaces for connectivity, leveraging artificial intelligence, so that you need much fewer sensors than what you would otherwise need to build out these maps for farms. And that’s been a concentration of the research teams that they are pursuing at Microsoft, in my team. And we are continuing to push the bar even lower (and) come up with new technologies to make it more and more affordable for the growers.
If growers want to use any of these technologies, I think there are solutions, and we are working with partners on building solutions that can be affordable for the growers, that can be at the price point the growers can afford. Because, even for us, the eventual question that technology providers and our partners are trying to address is: What is the price of technology such that the return on investment for the grower is much more than the amount that they are investing in these technologies? For example, through sensors or whatever, we want to provide insights that can help the farmer be much more profitable than the amount that they’re investing in deploying these technologies in the farm.
Tom: If any of our listeners would like to actually see this technology in action, is a demonstration available?
Dr. Chandra: Yes, Tom. There are few places where the people can see this in action. Of course, we are working — we have announced partnerships with other companies, which are starting to use this technology.
For example, we announced this partnership with the USDA, where — there’s a farm in Beltsville, Maryland, where we have deployed this. We also have a demo farm on the Microsoft campus; we do several demos here. And there’s actually another farm in eastern Washington. This is 9,000-acre farm spread over 45 miles. And there is a farmer I work with very closely, Andrew Nelson, he’s a fifth-generation wheat farmer, and he’s been using technology in his farm. He’s been using FarmBeats, he uses TV white spaces, (and) that is because his farm spreads over 45 miles, he doesn’t have to go everywhere every day. Using the connectivity, he can monitor where what’s going on.
He uses drones. He uses the drone’s information and combines that with leveraging FarmBeats. And then, he’s able to do the right application, the right intervention at the correct time in his farm. And there’s a video on the FarmBeats website where he talks about his experiences and how technology has benefitted what he’s doing in the farm.
And, in fact, his story just tells a very good story even otherwise that, as you probably know, one of the big problems in agriculture is the aging population of the farmers. The next generation of growers don’t want to get into farming. And, Andrew’s story — Andrew is, as I said, is a fifth-generation farmer. Like many others (in the) next generation of farmers, he came to Seattle, he did his undergrad in the University of Washington in computer science, and then he worked in the city for a while. And then, because of technology, he decided to go back to agriculture. And he’s going, he’s gone back, he’s farming again, (and) he’s using technology, he’s using all of the latest cutting-edge tech in his farm.
And this is a story which, as you can see, it appeals to many farmers of the next generation, the younger farmers, where you can then start using the latest in technology for your farm to see the benefits of how technology can help you farm better, help you produce more. And in the process, you actually use the latest and best in technology that’s out there.
Tom: You know, that brings me to a question I wanted to run by you anyway, and that’s how you’re using these technologies to have a societal impact.
Dr. Chandra: So, these technologies can help with sustainability. That’s one direction where we are actively working on leveraging these technologies, to estimate the amount of carbon that’s sequestered in soil. That is, using these technologies, a farmer can reduce their emissions because, you know — like, for example, they’re not using more chemicals than needed; they could be using good sustainable agriculture practices, like regenerative agriculture, like the distill. But not just that. Growers can use this technology to help increase the amount of carbon that’s sequestered in soil.
At the (same) time, when people have realized the importance of doing something for climate change, for reducing the amount of carbon emissions, agriculture can actually help provide the solution. Farmers can help put some of this carbon back into the soil. So, if you’re thinking of companies who want to, companies or organizations who are looking to reduce their carbon footprint, agriculture could provide the solution.
And using technology, farmers can use these regenerative agriculture methods while still staying profitable — not reducing their profitability — and yet, enabling a new income stream by putting carbon into soil. So, that’s one of the things that technology could enable.
In addition to that, we are also, at Microsoft, through Microsoft philanthropy, we are looking to bring technology to the rural population, enable the skilling of the rural population. For example, there is a skill gap, where there are quite a few jobs, but there’s not enough skilled population to fill those jobs.
And with FarmBeats, one of the things we’ve done is we’ve created FarmBeats student kits, we’ve created partnerships with the FFA with 4-H, where we are working with these organizations. And (we’re) working with, for example, the FFA chapters to bring technology into the curriculum of high school students even when they are in school, helping introduce them to the latest in technology, so that when they graduate, they are fully skilled in all of these technology methods, and they also know how would you apply these technologies in agriculture.
So, on the social good side, we are working on sustainability, on rural skilling, on air band, which is about providing connectivity to rural areas. We’re working on multiple directions to bring technology to the rural population in the U.S. and all over the world.
Tom: I wonder if the work with the FFA, the Future Farmers of America, is helping to overcome that reluctance of this emerging generation to go into the field of agriculture?
Dr. Chandra: I think so. And I think, as you show what’s possible with technology, more and more of the younger farmers will get excited to stay back in agriculture. In fact, when I talk to them, I tell them how, with agriculture, you’re seeing the latest in technology being applied to agriculture — with artificial intelligence, Internet of Things, with cloud computing, edge computing, all of that, the latest is coming to agriculture. And some of the FFA students that I have met, they’ve been so excited to be flying drones, to be using the latest in robotics in the farm. I think this exposure would help the next generation of farmers see the opportunities of what is possible by staying back in the farm.
Like, for example, the agriculture industry is not going away. We all still need food. And in fact, the importance of growing good food — the importance of growing food in a sustainable way — has never been more visible. Now, there is a bigger need for that, and technology can help address that.
And this is an opportunity: the more of the next-generation farmers, the more they see technology as a way to bridge that gap, the more entrepreneurial opportunities that exist. We see that more and more of the next-generation farmers would stay back, would want to, in fact, contribute — either leverage this technology or invent new technologies — to help agriculture be more sustainable, to help feed the world.
Tom: I don’t want to diminish or downplay the hard, hard work that is farming and the seriousness of farming and the things we’ve been talking about, but I have to say, what you’re talking about here sounds kind of fun.
Dr. Chandra: Yeah. And, of course, all of this is possible, as it builds on top of all the hard work that farmers do to ensure that all of us are getting good food.
Tom: What are the focuses of your current research?
Dr. Chandra: Right now, we are continuing to push the boundaries of technology for digital agriculture to make it even more affordable for the growers.
And one of the things — like, for example, we talked about how we’re looking to bring down the cost using artificial intelligence, using new connectivity technologies, but even then, the cost of a probe, a sensor probing the farm, is still a few hundred dollars. Like, for example, if you go look at sensors out there on the internet, it will be a few hundred (or) even, in some cases, a thousand dollars, which, while for some farmers here, it’s affordable, but a lot of farmers in the developing world, that still puts it out of their budget. They won’t spend like someone who’s farming an acre or a couple of acres; they won’t spend a few hundred dollars to put sensors in the farm.
And this is what we are doing with — one of the things we are continuing to investigate is: how do you bring down the cost of these sensors even more? How do you make sure that farmers can get data from the farm at an even lower cost?
One of the ideas we’ve come up with is to leverage Wi-Fi to send soil. Many farmers, they won’t spend a few hundred dollars to purchase a new sensor, but then they have a smart phone, even though it is an inexpensive smart phone. If they have a smart phone, it has Wi-Fi. And one of the new technologies we have built is where you can use the time of light of a Wi-Fi signal to estimate the soil moisture and soil electrical conductivity.
So, we’re envisioning a future where anyone can go to a farm, any farmer who has a phone can just bring their phone close to soil and can get the information about what’s happening in the soil, (or) can actually drive around, maybe, in a bicycle, and then you have a map of what the farm looks like.
So, these are things, this is just one example, but we are continuing to invent new technologies to significantly bring down the cost of data-driven agriculture. We want farmers everywhere in the world to be able to get information about what’s happening in their farm, things that they can see and things that they cannot see, at the price point which they can afford, so that once they get the data, you can then start bringing the benefits of artificial intelligence to every farmer in the world.
The vision here is to democratize technology, democratize data-driven agriculture, so that every farmer everywhere in the world can benefit from data and data-driven insight.
Tom: This has been so fascinating. Dr. Ranveer Chandra, the chief scientist at Microsoft Azure Global. Thank you so much for taking time for us.
Dr. Chandra: Thank you, Tom. Nice talking to you.