What’s next for robots in agtech? Everyone who’s paying attention knows agtech is massive right now … there’s so much innovation from laser-equipped weed killing machines that keep chemicals off our food and out of our land, to drones, autonomous tractors, AI-powered seeding and watering plans, and much more.
What about robotics?
In this episode of TechFirst, I chat about the future of robotics in agtech with Kevin Dowling, managing director at Robotics Factory in Pittsburgh, Pennsylvania. We discuss the evolution of robotics in farming, from traditional methods to the modern use of drones, autonomous tractors, and AI-driven systems, and Kevin highlights the diversity of robotic forms in agriculture, including wheeled, legged, flying, and swimming robots.
And — of course — we chat about humanoid robots with legs. Are they useful on farms?
Or … when will they be?
Kevin predicts a shift towards smaller, more affordable robots for smaller farms and emphasizes the importance of technology in reducing environmental impacts, enhancing food production efficiency, and potentially democratizing farming.
00:00 Exploring the Future of Robotics and Agtech
00:46 The Evolution and Future of Robotics in Agriculture
03:39 The Role of Humanoid Robots in Farming
07:38 Challenges and Opportunities in Ag Tech Startups
10:05 Innovative Startups Shaping the Future of Agriculture
12:49 The Complex Environment of Farm Robotics
15:30 The Potential of Indoor and Vertical Farming
23:30 Envisioning the Future of Farming with Robotics
(Subscribe to my YouTube channel)
Subscribe to the audio podcast
Find the podcasting platform you prefer:
Apple Podcasts
Spotify
Google Podcasts
Overcast
Other podcasting platforms
Robots in agtech: get the full transcript
John Koetsier: What is the future of robots in agtech?
Hello and welcome to TechFirst. My name is John Koetsier. Everyone who’s paying attention knows that agtech is massive. Right now. There’s so much innovation from laser equipped weed killing machines that keep chemicals off our food and out of our lens to drones, autonomous tractors, ai, it’s seeding and watering plans, all that and much more.
What about robotics? To chat, we have Kevin Dowling. He’s the managing director at Robotics Factory in Pittsburgh, Pennsylvania. He’s a former scientist at Carnegie Mellon Robotics Institute, serial entrepreneur, former VP Innovation for Phillips. Welcome, Kevin.
Kevin Dowling: Thank you. It’s good to be here. Thank you very much, John.
John Koetsier: It’s good to have you here, and maybe let’s just dive right into it. Is the future of robotics in Ag Tech, does it come on wheels, legs? Does it fly? Does it swim? All of the above.
Kevin Dowling: It’s definitely all of the above, and I think it’s ironic in that of course it started on legs with people and then animals.
Yes. And then it went wheels and tracks. So almost every form of locomotion there is maybe not serpentine, but certainly a lot of these others. So I think the future will be all of the above.
John Koetsier: Yeah, that reminds me, I think the word robot comes from the Czech and it means worker or something like that, right?
Yes. So yeah we have been our own robots sometimes compelled in history, sometimes not, fortunately, not so much anymore today, but I. Where do you see the most interesting innovation?
Because I’ve seen some cool stuff on wheels and that works in the farming environment and you’ve got massive, hundreds, thousands, tens of thousands of acres sometimes, and so big machines that can do big jobs there, there’s a role there, but there’s also probably some roles that, something that looks more like a human could do as well.
Kevin Dowling: Potentially, I think most of it because of the size and scale of the fields and also the equipment often either requires you carry a significant payload or that you gather a significant payload, tilling, seeding, harvesting, and the machine, like the human, would touch the field anywhere from a half a dozen to 30 times a year across the growing season.
And so you need to have machines that can handle that and today for the more automated aspects of farming them literally standing in their field. It’s tended to be row crops, the wheat, corn, soybeans, and so forth, where you have very large acreage of thousands of acres in a given farm field, especially in the Midwest here in the us.
And so that tends to drive that approach to very large machines that can cover 12 or more rows at a time. And that happens all the way through harvest when you have combines that do everything at the end. But what your fundamental question is what is the configuration, the morphology of these farming machines?
And it can be everything. Whatever’s best suited for that particular job will likely win out.
John Koetsier: Yeah, it’s also interesting because like you say, the big machines, the massive tractors, combines it makes sense given the scope and scale of North American farming. Maybe a little bit different in Europe, maybe a little bit different in Asia and other parts of the world as well.
But certainly in the States and in Canada, the massive farms there. It’s interesting. It’s interesting when we see this explosion of humanoid robots do you see that coming into farming in any way, shape or form?
Kevin Dowling: I think it’s interestingly anthropomorphic nature of humanoid robots is attractive because it’s what we are, what we resemble and what we look like.
How do you reconcile that with the job to be done. So, for example I’ll push back on that a little bit to say, if we were designing a dishwashing robot today, would it be humanoid and why? Where we have this very efficient box that we put dishes into, and that’s a robot for cleaning dishes.
So why would we bother with the humanoid form if we don’t need to do it? And airplanes don’t necessarily mimic what birds do. So although the humanoid is trending right now in terms of robotics investments and so forth, it’s not clear to me as it wasn’t with some of the walking dogs and other things that you see all over YouTube, for example, they are compelling because they look like things that we know.
And a part you have mentioned here the laser weed killing robot, which looks like something from space. And hopefully it won’t scale so big that it’ll become a threat, but it’s fascinating to see that I’m not convinced that a humanoid robot necessarily will be what you need in order to do tilling and seeding and harvesting and so forth. But I think for I will say though, that for smaller farms who cannot afford these large autonomous vehicles, the big autonomous tractors, there’s a real opportunity there because most farms, well, the US tends to much larger farms overall, but you mentioned other places around the world where these types of machines could be used on smaller farms, if they were affordable.
So, I point to a company, the big ag equipment makers John Deere leads the pack, but you also have Agco, CNH, but then rapidly rising behind them are companies like Kubota or Mahindra. Mahindra is now the world’s largest tractor manufacturer by unit numbers, not by revenue.
So they’re shipping more units to more farmers than anyone in the world. So where did that, where does that come from? What kind of technology can they add? What simple improvements and the computers and the sensors that are added to these machines are lowering in cost every time. And I’d add one, you pointed out that robot has derived from the check word for surf or worker.
The computer is the same way. The first computers were human. They were typically women in World War II who were actually calculating things all the time. And then that gave rise to the name, which was then applied to the inhuman, right. The vacuum tube based large computers of the past, and then became the the canonical name for calculating devices and obviously the computers we use today.
But there’s a lot of room for all kinds of machines.
John Koetsier: Absolutely. Absolutely. And that’s always the tough point, right? Like if you’re going to build a robot for washing dishes, hey, the dishwasher is pretty fricking good, right? Yeah.
On the other hand, it can’t load itself. On the other hand, it can’t empty itself.
Now, you could build in things for that, and the question is, do you build in, do you build 25 different types of robots that can do every task, or do you build one that can do what we do and make it work? And there’s no right answer or wrong answer to that, but certainly in farming it with the big machines that you need, you can totally see autonomous tractors and they’re here already to some extent, right?
Being a thing. I guess one of the challenges, and you’re in the business of finding, supporting growing startups in ag tech and related fields, that’s not cheap. If you look at the size and scale of those machines, that’s not the hardware Startups are notoriously challenging. I’m more used to software startups personally.
Right. But. When you get into Ag Tech and you look at the size and scale of these machines, the jobs they need to do they, the amount of funding required must be significant.
Kevin Dowling: It is, I think any hardware based company has a greater challenge than simply writing good code to solve a particular problem.
The advent of a large machine. But, even I’m not sure that the scale of the device matters so much as the ability to do the controls and everything. In other words, if you’re making, a new, let’s say like go back to the Nest or an Apple watch, very complex, small mechanisms, but the scale of that device matters less than the complexity of the thing you’re trying to put together.
So a tractor is typically an engine today. Sometimes you’ll find an electric motor and a drive, and those are well known, well understood, and made by. In the millions today across the planet. But to your point testing requires larger areas. You need real estate to do that. Sometimes it has to be seasonal, so you can only do it at certain times of the year, so that only compounds the challenges.
But there are many small companies focused on ag and doing it today, whether it’s mowing or harvesting or forestry all the way through row crops, specialty crops, and livestock not so much, but there is increasing automation even in livestock, beginning with milking parlors for dairy and, other things that people want to do.
I’ve talked to investors who want to invest in things that are heavily manual today, and the most heavily manual today. Still requires people to process animals into food. So, if you’re looking for a protein source like that there are many concerns around being able to manipulate and use the tools that they use in order to turn animals into food.
John Koetsier: What are some of the more interesting startups that you’re seeing that are coming across your desk and maybe that you’re investing in?
Kevin Dowling: So we there are about a half a dozen here in Pittsburgh right now. One is focused on vineyards and monitoring vineyards as well as doing harvesting and so forth.
One of the most interesting things that I have seen is that if you think about grapes, for example, normally if you just let them grow, they would become these scraggly sort of bushes, but they’re essentially trained. To wrap a structure and then form a particular thing. They’re now doing that with apples.
They’re doing that with other crops too, so that you can change what is in the field either potentially genetically. But the other way to do that is to simply provide a structure that they can grow in and around making it more accessible for automation, more easy to use. It also makes it easier, even for the human to do that, to reach the top of a cherry tree or an apple tree typically requires a specialized ladder or other equipment.
Or sometimes you can find these really interesting videos where they have shakers that come up and grab the trunk and shake the entire tree, and that’s how they harvest. So all of that I believe could be automated whether or not you use an, entrainment of a plant or you create plants that are genetically easier to harvest or produce an apple, for example, that is tougher and able to be shipped and stored and then eaten eventually.
So, there are companies like that. We have graduates of the programs here in robotics at Carnegie Mellon, for example, in Pittsburgh, who are off doing these types of things elsewhere. We also have a company you mentioned earlier the idea of indoor agriculture. The idea is you have more than a greenhouse where it’s actually set up to be fully automated. And we have a company here for growers, which is harvesting tomatoes in greenhouses. And they’re doing this in, in, not only here in the US but in the Netherlands, which is a hotspot of of indoor agriculture as well as, excuse me, as well as automated agriculture.
And they’re harvesting tomatoes rapidly.
John Koetsier: That’s really interesting because a lot of tech comes out of the Netherlands because they’ve vastly expanded their agriculture on a tiny land footprint. So they’ve gotten very sophisticated at getting high yields and high automation. ’cause their costs are high and everything like that.
Obviously in flowers, but also in agriculture. That’s right. Super interesting that they’re taking their tech to the Netherlands.
Farms are a crazy challenging environment to put robotics into. Right? They’re complex. They’re changing, they’re moving. It’s not laid out like a factory. It’s not relatively stable. Ground conditions change. There’s weather there’s animals, right? And every type of farm is different as well. A wheat farm is very different from a vineyard, as you mentioned or, a beef dairy concern, that sort of thing.
That leads to significant challenges, I’m assuming.
Kevin Dowling: It does. It’s not entirely unstructured. It’s not a walking into the woods, for example, where you have no control of anything. But it is even though it’s. It is slightly structured, it’s still challenging. It’s soft materials that you have to traverse. You’re also some big concerns right now in agriculture have to do with soil compaction.
So how do you prevent that from happening as time after time as these machines roam these fields? How do you prevent those kinds of things? Well, you can make them bigger, have softer tires, have more tires. More legs perhaps depending on the approach. So, and the farmer is assuming most.
In some cases, I think people are arguing all of the risk because they face those challenges. It’s outdoors and you’re subject to all of the vagaries of of the outdoors in general. And the, so they get the harvest time and they produce the crop. But if it’s, if they had bad weather, if they had floods, if they had drought they’re the ones assuming the risks.
So how do we help the farmer to mitigate that risk? One way as I mentioned, is moving indoors. But who knows? There, there can be other potential ways. There are crops, for example, that can be grown entirely indoors like mushrooms. And you might, you and others might be amused by the fact that mushrooms are a crop, but they are, and they harvest every several weeks.
It is actually Pennsylvania’s number one crop by volume, by dollars. It’s, wow, six to $700 million a year to the state of Pennsylvania. Wow. So it’s a, and if Pennsylvania was a country, it would be number four in the world in the production of mushrooms. So it’s very likely the mushrooms you have on your pizza or in your salad are made right here in Pennsylvania.
But they have problems harvesting because of labor issues. So how do you solve those?
John Koetsier: Yeah. Yeah. It’s interesting that we had the discussion at the top, like, what is a robot? What do you classify as a robot and what’s the most efficient sort of machine, whether you call it a robot or not, to operate on a farm.
We talked about bringing farms indoors. Vertical farms are a thing. Right. And there’s huge potential there for year round production. Right. There’s huge potential there for production where. It’s needed, so right in the city, let’s say, or nearby or production for, let’s say far northern climates, Alaska’s the Alaska’s of the world, right, where farming is not really a thing. It’s a thing, but it’s not really a thing. But fresh fruit and vegetables are required. In that case the farm itself is a giant robot.
That’s an interesting world to consider too.
Kevin Dowling: Yeah it is we have a rather broad definition of what robotics is, and I think of it as the cycle of sensing planning and actuation.
And so you take that loop, that triumvirate of these three aspects of it. One is capturing the data planning based on the data you’ve acquired, turning that into information and then turning that into potentially movement. It doesn’t have to be movement, but a thermostat in your home also has follows that same cycle these days.
It’s not necessarily a robot, but you, to your point John the idea of a home, I think a famous architect, Le Corbusier said, a home is a machine for living. And I think we’re living inside of robots because our climate is controlled, our temperature, humidity many other things about our home, the lighting everything can be controlled.
And so that’s an environment that we’ve created for ourselves. That is robotic in nature. But agriculture I think could be exactly the same type of cycle, the exact same type of thing where we’re trying to control things in order to produce food or do some task in a factory, for example.
But all of those loops together, so I think your analogy of the farm being a robot is exactly spot on.
John Koetsier: Is there progress or is there, are there even projects to create something like a FarmOS? Because let’s say, we’re five years in the future, 10 years in the future, and you’re starting to see a significant surge of. Automation and robots and semi-autonomous machinery on farms, there’s a command and control issue, or there’s at least a work together issue.
Right. And do these things know about each other and do some of them cooperate and some of them are harvesting, but then some are gathering and transporting and do they talk and how’s that communication? Is there a project underway to, to manage that or is it basically using. Some of the operating systems that people are developing for coordination, robotics in general.
Kevin Dowling: I, I think most of it is exactly your last point, which is using real-time operating systems, for example, or using Ross Robot operating system to control the robot itself. But in terms of complex networks of robots, there’s not a whole lot of work going on in that area. I do think it will be absolutely necessary to do that if you have.
I believe that there will be a trend to move away from the very, very large machines, the half million dollar 12 row kinds of systems to larger numbers of small machines, which are more easily maintainable. You can pull them out if you need to. And by having a larger number of smaller machines means that they do have to communicate, they do have to know where each other is.
And that will require, as you pointed out, the ability to communicate amongst themselves like a mesh network in a way, but mobile and being able to coordinate and synchronize their operations. If you think about in a field today, even the. The operation of a harvesting, let’s say you’re moving through a field of corn.
The US has around 90 million acres of corn, so there’s a lot of it. But as you’re harvesting threshing as you’re generating the grain and the cobs and so forth that are becoming eventually silage or other things, there, there’s a coordination of vehicles, one driving to capture that flood of granular material coming into one vehicle from another.
And so there’s already a little bit of that work. But typically one and twos not tens and twenties of these machines working together. sometimes you’ll see these wonderful, beautiful pictures of probably Kansas or further north where they’re row after row of combines working together.
Yes. And then they migrate …. those aren’t robots yet, but they essentially migrate northward with the harvest. And that’s a specialty machine that allows farmers to harvest a great deal more quickly. And because they can’t afford that many machines, no one really can. Yeah. But as they move north toward Canada it allows them to do at a, do it at a rate that is far faster than anyone could do in history prior to that.
John Koetsier: Interesting. Super interesting robotic migration. Here we come. Very cool. Do you do any investing in, I wanna say farming adjacent areas? Because there’s a lot of work, let’s say we don’t wanna have however many tens of millions of cows that there are in the United States. States and Canada and we don’t want to kill that many for food.
And so we want to grow beef and have the impossible burger and other stuff like that. Lab grown meats and other things like that. Do you do any investing in those areas?
Kevin Dowling: We haven’t. I think that’s probably a little beyond our direct knowledge and capability. I have nothing against that. I think creating protein sources like meats from sort of chemicals is certainly a valid way to do that and reduce some of the impacts that the the full animals have.
But I’m not especially going after one versus the other. What perhaps, one way to think of this is there’s the farm and that’s really what we focused on in our recent Ag tech summit, which is specialty crops and row crops. But we did ignore for the purpose of this particular summit, livestock, pigs, chickens, cows and so forth. Cattle. And then the beginning there’s also seed. Chem fertilizer, chemicals and other things. And there’s a, there are ways now there’s one company here called Robot, and it’s spelled R-O-W-B-O-T. They’re focused on small robots going between rows of corn, which happened to be planted perfectly on center, 30 inches apart.
And so they designed a robot to do exactly that and inject nitrogen directly at the roots of the plants. So it cut the use of nitrogen, which has other impacts in the environment of course. It ends up in the Mississippi River. It will then turn, make its way to the Gulf of Mexico and so forth.
But it cut the use of the nitrogen fertilizer by half
John Koetsier: Nice. Which is,
Kevin Dowling: Not only a great cost savings, but reduces that impact on the environment substantially. And they’ve been working for quite a bit and hundreds of acres in Iowa right now.
John Koetsier: Absolutely huge. If you look at the amounts that farmers spend on fertilizer, it, you have, obviously, it blows your mind.
There’s huge costs here. And like you say, and nobody wants all that flowing into the lake. That’s right. The analogy that comes and everything. Right. So, yeah, exactly. Okay. Let’s look forward what does farming look like in, let’s say 10 years? As we see more. Purpose. Robotics and machinery.
Autonomous, autonomous machinery for the farm. What does farming start to look like?
Kevin Dowling: I think it’s certainly an extension or an extrapolation of what we see today. There’ll be more machines and more fields doing these types of operations at the beginning. Of course, there’ll be the much larger farms which can afford this machinery or perhaps there’ll be some interesting, ways to financially incentivize the farmers to lease these things, the capital lease, operating lease all kinds of financial ways to manage that to further it. And then I think you’ll begin to see taking that same technology which will get cheaper, but once, once it’s become. Solid and robust in terms of use.
You’ll start to see that in smaller farms where if you have not 20,000 acres but 2000 acres, you can also make use of that. I don’t know how much smaller it will get if you have 20 acres or two acres of farm. Will it have. Autonomous vehicles. Yes. Eventually, but, and you asked about the next five to 10 years.
I’m not sure that it will, but I hope that it will. And so that, those are the field operations, and then there could be things related to fertilizing and inspection. It could be drones and other things to monitor crops, look for disease basically remove those that are diseased from the field.
And all of that could be done in an automated fashion. What’s interesting to me too, a lot of what we learned was post farm. How did, when you go from harvest into the processing chain, there were a lot of interesting, both economics and potential automation. So, for example, we’ve seen many across the country, I’m sure where you are and certainly where we are in Pittsburgh.
The rise of many small breweries. But very often they can’t scale well because for a small amount you have the machinery and the existing industrial tools to be able to satisfy a brewery’s output, which might be, who knows, thousands, a thousand cases a year, something like that. But when you start to scale beyond that, there’s a gap actually in the market between what a Nestle can do or a very large food company can do, and what a small growing.
Bakery or other things. So there’s an interesting gap in there. We’ve been talking to contract manufacturers who work in that space about how to improve that. And there may be room for another summit related to post-farm activities and food processing.
John Koetsier: So my hope is also that in whether it’s 5, 10, 15 years, as this becomes more available to farmers, this technology we also have a better product.
Yes, we also have better food. We use less fertilizer because it’s, as you mentioned, injected right where it needs to be, not just carpet bombed over the farm. Right, right. We use less pesticides because we have machines that either mechanically or with lasers, remove them stuff that we’re seeing right now but isn’t necessarily widespread.
Right. And maybe we have less waste because we’re smarter about how we harvest. Maybe we lose less to crop failure because as you said, there’s constant monitoring, perhaps drone based or something like that so that we can identify, oh there’s. There’s something attacking the corn here.
Let’s, get rid of that. Let’s treat that right away and it won’t spread. So my hope is that farming becomes safer for humans, but also better for the environment and also better for people who.
Kevin Dowling: Yes I certainly have the same hopes and dreams around that. Organic farming, for example.
There may be a lot of ways where machines can help grow, not, no pun intended, to grow that part of the market as well. Where we could have healthier food that’s grown. And here’s another aspect of it, is, everybody sees farmer’s markets. You might see the honest stands where you’re, you basically put money in a jar and you take a, a dozen years of corn.
There might be ways to do monitoring of that to reduce the need for people to be there all the time. Mm-Hmm. And then you have companies like one here in Pittsburgh called Harvey, H-A-R-V-I-E where they’re going directly from the farmer to the customers. So they’re able to navigate, they work with something like 300 to 400 different farms.
So that’s another link in the chain of being able to provide good food for people on a timely basis where they can directly they’re essentially directly paying the farmer. There might be a middleman in there doing some of the distribution, but, I think there’s I think the, I, hesitate to use this term since it’s used so often, but the democratization of farming and how to get it so that the farmer can benefit more directly from it without a lot of processing and other things going on in between them and the consumer.
It’s really opened my eyes to see all of this. It’s amazing.
John Koetsier: It is amazing. And like you, I also hope that these innovations will come down to the smaller farms because we’ve seen some interesting, amazing, incredible things in small farms right now. Even as small as one acre on a city lot or something like that.
Where with intensive farming techniques and there, there’s really a lot that can be done. Wanna thank you for taking this time. Do appreciate it.
Kevin Dowling: You’re welcome. Thank you, John.
TechFirst is about smart matter … drones, AI, robots, and other cutting-edge tech
Made it all the way down here? Wow!
The TechFirst with John Koetsier podcast is about tech that is changing the world, including wearable tech, and innovators who are shaping the future. Guests include former Apple CEO John Scully. The head of Facebook gaming. Amazon’s head of robotics. GitHub’s CTO. Twitter’s chief information security officer, and much more. Scientists inventing smart contact lenses. Startup entrepreneurs. Google executives. Former Microsoft CTO Nathan Myhrvold. And much, much more.