Host: Dr. Ping Yu, UGA Department of Horticulture Guests: Dr. James “Jim” Robbins, Extension Specialist and Professor (retired), University of Arkansas Division of Agriculture; Dr. Joe Maja, Director, Center for Applied AI for Sustainable Agriculture, South Carolina State University Producer: Rich Braman Episode: S02E04 — Blooms and Beyond, Season 2 (special two-parter, Part 1) Duration: ~66:20 Air Date: June 21st, 2026
Welcome and Introductions
Ping Yu: Welcome to the Blooms and Beyond podcast, a podcast that uncovers plant health, history, culture, and management through the lens of science. I’m your host, Ping. How’s everyone doing today? I am doing great, ‘cause I’m about to talk to one of my favorite people here and one of the most exciting topics on today’s episode: drone application in the nursery and greenhouse, with Dr. James Robbins and Joe Maja. So I don’t wanna steal any more thunder from our speaker today. Without further ado, here are the conversations with Dr. Joe Maja and James Robbins. I hope you enjoy.
Good morning, Doctors James Robbins and Joe Maja. Welcome to the podcast. But first, let’s start off with the introduction. Can you guys tell us a little bit about who you are and what you do?
Jim Robbins: Yeah, I’ll jump in first. Thanks, Ping, for the invitation. So, Jim Robbins. I’m a retired extension specialist and professor in commercial ornamental horticulture at the University of Arkansas Division of Agriculture. For over 23 years with the University of Arkansas, I provided support for green industry businesses across the state, focusing on nursery production and woody plant evaluation. And as we’ll get into a deep dive, in about the last 15 years, I got very heavily involved with drones, thanks to my collaborative friendship and interaction with Dr. Joe Maja, who’s our second speaker. Joe, do you wanna take over from here?
Joe Maja: Hi, I’m Joe Maja. I’m currently the director of the Center for Applied AI for Sustainable Agriculture at South Carolina State University. My background is in engineering, with a focus on robotics, unmanned aerial systems, remote sensing, and more recently, integrating AI into agricultural applications.
Over the years, my work has evolved beyond just drones. We’ve been working on systems that combine UAVs and ground-based robotics, RFID technologies, and data analytics platforms to address real-world challenges in ag, particularly in nursery and specialty crop production. So a big part of what I do is taking advanced engineering concepts and translating them into something practical and usable for growers. And that’s really where the collaboration with Jim has been so valuable. He brings deep expertise in horticulture and production systems, and I bring the engineering and systems integration side. So together, we’ve been able to develop solutions that are not only technically sound, but also meaningful and applicable in real operations.
Ping Yu: Okay, thank you. So it’s kind of interesting for today’s episode, because normally we would have only one expert to talk about a certain topic, but today our two speakers, they come from very different backgrounds.
From Plant-Lover to Drone Pioneer: Jim’s Journey and a Serendipitous Collaboration
Ping Yu: But can you tell us a little bit about — for Dr. James Robbins, can you tell us a little bit about how did you start with horticulture? Is this something interesting to you when you were a child, and how did you end up — can you walk us through your career journey in horticulture in general?
Wisconsin nurseries, Michael Dirr, and a unique mix of industry and academia
Jim Robbins: Yeah, I’ll give a brief background on that. So I was raised in Wisconsin, started working at a nursery — a really prominent nursery and garden center in Wisconsin — probably at age fifteen, and just really knew that I loved plants. I was fortunate; I went to the University of Wisconsin, Madison, had an exceptional undergrad advisor that really recognized my enthusiasm for plants. Then I was very fortunate to go on and get a master’s with Michael Dirr in his first few years at the University of Georgia, and at that time he was also the director of the State Botanical Garden. And so each of the new grad students, we had a role there.
Then I went out — but it was just, you know, I look back and I’ve been blessed with great, prominent mentors to just continue feeding my interest in plants. I went out to UC Davis, where I got my PhD in plant physiology. And then over my professional career, the breakdown is probably like — I had a unique mix: nine years in industry, and then the remainder, whatever it is, twenty-three years, in academia. So kind of a unique opportunity to be on both sides of the fence.
The 2009 question that started fifteen years of drone work
Jim Robbins: Real quickly — I think it’s very important, ‘cause you mentioned it, it’s kinda unique that you’ve got two speakers and they’re from two different disciplines. I really wanna point this out, because I’m very fortunate, and the industry is fortunate. Back in 2009, I was at a very prominent field nursery in the Pacific Northwest, and they had raised a question about, gosh, was I aware of a way to automate inventory for nurseries? And I was just shocked that we really didn’t have, beyond manual counting, which is really the prominent method.
So on my return to the University of Arkansas, I talked to my bio and ag engineering colleague, Dharmendra Saraswat, who had just come on board from the University of Florida, and his prior major professor, Reza Ehsani. They had just completed some work looking at inventorying, or looking at parameters of citrus groves, so they could count them. They could also measure the canopy volume, which would help them predict yield. And so at the time, Joe was there, and we all got together, with maybe no funding. So this is back in 2009. We went out to this nursery, and we’re using a ground-based method to try to improve inventory in field nurseries. And so that’s the first time I really met Joe. We just really hit it off. He’s just brilliant, and just a great human being.
So then it was — for a variety of reasons, we dropped the ground method in spring of 2010. So think about that, it’s fifteen years ago. Joe had kind of caught the early whiff of these drones being used. And these — they were hobbies at this time. People can’t appreciate — if you’re jumping into this today, you’re saying, “Gosh, this is a piece of cake. These are easy to fly, they’re inexpensive. The flight navigation software is wonderful.” But that really didn’t exist in 2010. These were mostly built kits, and Joe built quite a few for our various projects. They were difficult to fly. We didn’t have the fancy flight navigation software. But we jumped into this with the thought that maybe we could use drones in this plant inventory process. And that really was for any open field application — that was the thought at that time.
And so over the years, we’ve worked on that. I had a great master’s student, Josue Nahum Leiva, who did his master’s project looking at analyzing aerial images for counting. And that is one approach that can be used. There’s a number of companies that offer that service — actually, it’s called plant counting. So that is available. But then we kind of evolved into a collaboration with Tom Fernandez at Michigan State, who is interested in RFID tags, and that kind of transitioned into our recent grant-funded project, which is to try to use RFID tags and either a handheld or a drone-based reader to improve, or automate, inventory for nurseries or for greenhouses.
And so it’s been an evolution over time. And Joe has been a key component. I’m not an engineer, you know. And so I encourage others that are involved in these kinds of projects — I mean, I help Joe kind of understand the production issues, and that helps feed into his improving his designs of engineering solutions. So it’s been a wonderful collaboration, and it’s been necessary, because we kind of need both components to provide a successful solution. So with that, I’ll stop and turn it over to Joe. But again, I want to point out that this has been, for me, just a blessing to have this serendipitous run-in with Joe back in 2009. And then just the realization that we could become a rather good collaborative team and friends over this many years. So Joe, over to you.
Joe’s view: experimentation, scalability, and the aerial perspective
Joe Maja: Yes. From my perspective, those early years around 2009 and 2010 were really about experimentation and trying to solve a very specific problem: inventory management in nurseries. So at that time, we initially explored ground-based solutions, thinking that mobile systems could navigate through rows and track plants. But it didn’t take long to realize that scalability was a major limitation. The environment was too complex, and the effort required just didn’t match the efficiency we were hoping to achieve.
So around the same time, I started seeing early drone platforms. These were not the commercial systems we have today, but more like hobbyist kits. Still, it was clear that the aerial perspective offered a completely different way to approach the problem. So what made the collaboration work so well was that Jim helped me understand the operational challenges from a grower’s perspective, and I could then think about how to design systems that actually address those challenges in a scalable way.
Why interdisciplinary collaboration matters
Ping Yu: Yeah, ’cause I think you guys are a perfect example of how important, or how unique, the collaboration can be in terms of your professional development and friendship. I know you guys for a long time, and I know the relationship between you two is something unique. And right now in agriculture in general, it needs multi-discipline collaboration to make things work, and you guys are a perfect example for that.
But Joe, can I ask you this — do you know plant, or will you say you’re a plant person before you met James?
Joe Maja: Honestly, no. So I wasn’t a plant person before working with James, and my training was purely in engineering, so I approached everything from a systems and problem-solving perspective. But through the collaboration, I gained a much deeper appreciation for how complex and dynamic nursery systems really are. So it’s not just about plants — it’s about logistics, timing, environmental conditions, and operational efficiency. So in many ways, it became a true interdisciplinary experience. Jim helped me understand the plant and production side, and I helped bring in the technology, and that combination is really what made the work successful.
Ping Yu: See, that’s the unique part — James kind of introduced you to the plant world, while you kind of drag him into the more fun drone project. So that’s very unique. That’s what I see here.
The outdoor environment is rough on engineered solutions
Jim Robbins: Yeah, and I kind of dragged Joe — [laughs] probably unwillingly — into this. I kept bringing up ideas of projects. Again, because of his expertise in remote sensing and robotics — I mean, he does a lot of work on ground-based unmanned vehicles and aerial vehicles. But his breadth of knowledge and skill set is just tremendous. So we talk about the problems, and then he’s just been phenomenal in coming up with workable solutions.
And I want to emphasize, since the majority of our work is in outdoor nursery production, it’s very challenging, and I think Joe would agree with that. [laughs] You know, we talk about, like — if you wanted to get RFID and inventory in a heating/cooled warehouse with a concrete floor and racks that are very uniform and all that kind of stuff, that’s well-researched. But boy, when we start running equipment out in agriculture, where you’ve got dust and dirt and wind and rain and sun, it’s the transfer of knowledge that is not as easy.
And so I think people in the industry appreciate that. If you think of the harvest robotics and the container-moving device and the number of iterations where they needed to look at the difference in container bed layouts and the types of materials that are used — the outdoor environment is very rough on engineered solutions. And so hopefully people understand and appreciate that it’s definitely much more difficult than, say, in a controlled-environment situation.
Ping Yu: Yeah, I agree with that. And I want to thank you guys for your hard work, because what I remember when I was going through some of my slides for one of the classes that I’m teaching — there was one slide with you, Dr. James Robbins. Your quote, I think it was based on one of the projects that you had done back then, in maybe 2010, 2012, about your prediction for the future in nursery production, was like: automation would be one of the pioneering methods or tools for our industry to kind of evolve.
And it’s true right now, because over the past couple years — I don’t know if you guys have noticed that — but here in Georgia, and I know in other states, not just in the institution perspective, we have hired a lot of people from precision ag, people who emphasize more in engineering and focusing on automation and advanced technology application in agriculture in general. So I’ve seen the trend of adapting the more advanced technology into agriculture, which is great, because I think one of the top issues everywhere you go — people will say labor is a top issue. And using those advanced technologies in agriculture would be one of the tools to address the issue.
Introducing UAVs: the 55-pound line, and multi-rotor vs. fixed-wing
Ping Yu: But before we jump into it, can you guys give us, for people out there who don’t know a whole lot of that advanced technology in agriculture — can you just give us a brief introduction of the UAVs and what types of drones are out there? I know people, there are personal uses for drones, just for the hobbies and taking pictures and doing vlogs. But for the commercial use in agriculture in general, I know a lot of nursery and greenhouse growers are trying to adapt more. Can you just give us a little bit about the big picture out there?
Jim Robbins: Yeah. And I think the other thing — there’s this terminology, which has evolved. Joe and I can, again, because we were really at the earliest stages of the explosion of these drones, or UAVs, UAS, whatever you want to call it — in our group, we actually had to come up with several names at that time, because we didn’t even know how to describe it. And we had a meeting with the FAA early on, and they said, “Oh, you can’t use that word.” And so it’s interesting, the nomenclature — even that was not, 15 years ago, hard and fast. It was still a debated point. So it’s just interesting in 15 years to see the evolution over time.
But one thing is, there are these two names: UAV, unmanned aerial vehicle, or UAS, unmanned aerial system. So if you’ve got a sprayer, or the payload is a sensor, you might want to call it a system, I guess. If you’re just flying the aircraft, it could just be unmanned aerial vehicle.
And then there are recognized, basically, two categories. So a number of years ago — back, I think it’s 2016 — the FAA finally came up with some very defined regulations related to UAVs, and so they defined something called the small, so small UAV, small UAS. And what that meant was, when the unit takes off the ground — so it’s the aircraft and the payload — if you were to put that on a balance and it weighed less than 55 pounds, it fits into a category called small. And so it’s important, because a lot of the early regulations, there was a lot of detail related to that, and getting licensed in an sUAS application was very achievable.
It doesn’t mean that you can’t fly a drone that weighs more than 55 pounds. It’s just a more difficult regulation. So think of your driver’s license — just a regular driver’s license. But if you wanted to be a certified CDL, it’s a much more expensive and involved process. And the same is true once we jump above 55 pounds. So that’s important for people to know. And if you’re gonna get into spraying, pretty quickly you’re probably gonna jump into that category. So it just means — it doesn’t mean you can’t fly, it just means that when you start looking at the regulations, you’re gonna need to focus on this specific area and not the sUAS rules.
So that definition of that category of small aircraft was useful, because it kind of gave us something firm to work with, and a lot of flexibility. But like I said, today, with the increased interest in aerial application — or we’ll call them spray drones or spreader drones — a lot of these pretty quickly are gonna default, fall over. At takeoff, they’re gonna weigh more, likely, than 55 pounds. So that regulatory thing is just gonna be a little more involved to process.
Joe Maja: Yes. When we talk about UAVs or drones in agriculture, I usually break them down into two main categories: so, multi-rotor and fixed-wing systems. So multi-rotor drones, like quadcopters or hexacopters, are the most common, and they can hover, they’re highly maneuverable, and they’re well-suited for operations where precision is important. So in nursery environments, where you have tighter space and more variability, multi-rotors tend to be the preferred option.
Now, fixed-wing, on the other hand, are designed more like airplanes, so they’re efficient over large areas and can cover more ground in a single flight. But they require more space to operate and don’t offer the same level of control in confined environments. Another factor to consider is payload capacity and flight time. So multi-rotors typically have shorter flight times, but can carry specialized sensors or application systems, while fixed-wing platforms prioritize endurance. So at the end of the day, the choice shouldn’t be driven by the platform itself. It should be driven by the application. Whether it’s imaging, monitoring, or spraying, the key question is: what problems are you trying to solve?
Jim Robbins: So I think that’s very important for people to understand, that there’s this kind of two categories.
Ping Yu: Yeah. I like your analogy, your example of comparing flying a drone versus a driver’s license, because basically, if you have a driver’s license, you can just drive your own vehicle. But if you wanna be a bus driver, or a commercial driver, or a truck driver, you probably need to advance your driver’s license test.
Jim Robbins: And this designation, in this case, is simply based on a weight issue. So it’s important for people to know that that 55 pounds is something you need to just have floating in the back of your mind.
Licensing: Part 107, Part 137, and Who Regulates What
Ping Yu: Basically speaking, since we’re talking about the license — what are the licenses that are needed, for instance, for the small drone, or a personal one?
Jim Robbins: So again, back in — I think I’m correct in saying it was back in 2016 — when they rolled out these formal rules, which was very important, because prior to that, it was kind of the Wild West. And so you need — as these technologies advance, this is very normal — the technology advances faster than the regulations can keep up. And so finally, they did come out with these rules, and again, I believe it was 2016.
And at the time, they defined three groups of users: public — so that applied to Joe and I, with the universities — commercial, and what was it called, Joe? Is it private? It wasn’t consumer. The hobby group. So there were three buckets at that time. Over time, they’ve kind of dissolved that. And now we’re, I think, down to two, I would generally describe.
But there’s this Part 107, which is — so more than likely, if you’re gonna use this — it’s your nursery or greenhouse — you’re gonna need to look up Part 107 and become a certified pilot. And the process, again, over time, has gotten much easier. So it’s still something that you need to complete, not really overburden. And let’s probably not even get into just the hobbyist use. Even today, they’ve gotta take a minor test just to get certified, and then where they can fly is controlled. But really, the purpose of this podcast should be for people that are gonna be more for this commercial use. So Part 107 is something that, again, you need to write down.
It doesn’t matter who you are, whether you’re a hobbyist or commercial — you’ve gotta register your drone. And that today is a very, very easy online process. Just a couple minutes to do it. Very inexpensive. And again, we don’t need to get into the weeds on it, but compared to when Joe and I did it, it was a paper process, very, very difficult. So things, again, over time, have improved for everyone.
And then the other licensing — let’s say you do have interest in an aerial application — then you’re gonna need to also look at the FAA Part 137. And this is necessary for using drones for applying pesticides or agricultural chemicals. Now, today it’s again gotten a little easier, ‘cause this is kind of a new category, probably two years old. And so you could do it on your own, or there are people that you can pay to help you go through. But there’s a lot of detailed paperwork involved in going through this process. So if you’re not going to apply any aerial chemicals, then really it’s registering your drone and Part 107 that you really need to focus on.
Joe Maja: Yeah, so from a practical standpoint, Part 107 certification is very achievable for most people. So even without an aviation background, the exam focuses on topics like airspace classification, weather regulations, and operational safety. Where people tend to struggle is understanding how airspace works and how to operate legally in different environments. That’s usually the biggest hurdle. But with some focused study, it’s manageable.
So Part 137, on the other hand, which is required for aerial application, is more complex. So there’s additional regulatory oversight, more paperwork, and in many cases it involves working with state agencies as well. So for someone starting out, I usually recommend focusing on Part 107 first, gaining experience, and then exploring Part 137 if they plan to get into spraying operations. And in some cases, it’s worth working with consultants or experienced providers to navigate that process more efficiently.
A Field Guide to Drone Types and Sensors
Ping Yu: But can you give us a little bit more about what types of drones are out there for people to use for different purposes? Can you give us a general introduction to different types of drones?
Jim Robbins: Joe, you want me to start, and then you can add?
Joe Maja: So when I think about drone application in nurseries, I group them into a few categories: you know, imaging and mapping, crop monitoring, spraying, spreading, and inventory or counting systems. So with imaging, you can have standard RGB cameras, multispectral sensors, thermal sensors, hyperspectral cameras, and even LiDAR systems, depending on the level of detail needed.
So one new thing I’d add is that these categories are becoming less distinct. We’re seeing more integrated systems where a single drone platform can support multiple applications depending on the payload and software. So instead of thinking in isolated categories, it’s becoming more about flexible systems that can adapt to different tasks.
Ping Yu: So right now, people are kind of merging the gaps in between categories, and so one would actually function a couple different functions.
Jim Robbins: And one more, Joe — just general photographs. I mean, I guess we could put that under the imaging. Yeah, just simple — people wanna get video or photographs for, say, marketing or sales, something.
Ping Yu: So basically, based on the purpose of the drone application, we can categorize them into general imaging and plant nutrition monitoring, spray drones, or spreaders for granular application, and a counting system. Basically, five or six of those categories. For all those different purposes of drone application, are the requirements for the license similar, or are there different requirements for different applications or different purposes of using the drone? And who’s the agency they have to work with — just the FAA, or are there any other agencies involved in this whole process?
Jim Robbins: In simple terms, Ping, anything that deals with flying your aircraft, that is the sole responsibility of the FAA. But if you are gonna apply some kind of a chemical, then, for example, in the state of Arkansas, the agriculture department, the pest control division — you may need to have a conversation with them. So, but anything that deals with flying an aircraft in the national airspace, which is what we’re talking about, is the sole responsibility of the FAA.
Ping Yu: Okay. And so basically, they have to go through all those paperworks to get their licensing, and depending on what purpose they are using their drone for — for spray chemicals, for counting — they may or may not have to go through additional steps to finish the whole process.
The Five Drone Applications — and Where Nurseries Use Them Most
Ping Yu: But with all those different types of drone application, have you guys seen, over the years, what would be the most — I guess — most used drone application? ’Cause I know right now, due to the safety, or employee safety purpose, a lot of people are trying to use the spray drone a little bit more. But have you seen the same trend over the years, where people are actually using drones basically anywhere they can?
Marketing and asset tracking
Jim Robbins: Well, so, Joe and I have stuck to a list of five kinda applications that we set out a number of years ago. And I’m sure that somebody could adjust this, but here they are.
So the first one is for marketing and sales. And so this would be using your drone for high-quality imagery that enhances marketing materials. And any operation — I don’t care how big or small — should have a small unit, probably gonna cost no more than five hundred dollars. I mean, you could certainly get something for less than that, but that is durable for that purpose. ’Cause I think we all now recognize the importance of aerial images or aerial videos, again, in sales and marketing.
The second is something that we had defined as asset tracking and management. And that would be, for example — let’s say you’re a large grower, and I’d like to know, how big is my bark pile? So how much do I have sitting there, so I can estimate how many containers I can fill? So today, using some image processing methods combined with an aerial image, we can estimate that quite accurately. So that’s a good example. Or let’s say irrigation coverage — or if you’re a greenhouse, and you’ve got a retractable roof, or you’ve got gutter-connected houses, and you need to inspect those structures safely, it’s really the only way to go. And again, this is that same three-, five-hundred-dollar unit.
Plant inventory and RFID
Jim Robbins: The third category that we’ve talked about is plant inventory, and that’s the one that right now Joe and I tend to be most focused on. And that could either be — over time, again, we’ve looked at both image analysis for counting, but more recently we’re laser-focused on using RFID tags combined with either a ground-based handheld unit, or — Joe is designing some phenomenal, very small, lightweight aerial reading unit that you would attach to a drone.
Aerial application: spraying and spreading
Jim Robbins: The fourth category is this aerial application — so you could call it spraying. I like to emphasize that it could also be spreading. So Ping, I mean, in Georgia, fire ants is a big issue, and so, fire ant bait application. And really this is an exceptional application of a drone, because you can lay out the pattern that you want to apply, and since it’s bait, it doesn’t need to be overlapping. So you can design this course very easily on your screen. And because these materials tend to be a lower bulk volume, you can carry quite a bit of it. And so you can apply these materials probably the most efficient way I can think of. So in the south, southeast, that’s a great —
And the credit goes to these aircraft designers. In just the last three years, they’ve done a remarkable job. It used to just be a sprayer with a liquid tank, and then very easily they design where you can pull the liquid tank out and swap out an interchangeable granular dry tank with a cyclone spreader. It’s truly amazing how easily you can adapt from a liquid system to a granular system.
Crop monitoring
Jim Robbins: And the last thing is crop monitoring. And Joe has referred to that several times. So that — whether we’re looking for water stress, which Joe and I have done some work on, heat stress, weed infestations, chemical damage — all of those, using remote sensing to try to identify these areas before really the human eye can pick up on it. And that’s not new. We’ve been doing this, but with low-flying manned aircraft. The unique area is that we’re now using these small, very maneuverable drones to do it. And the key to the whole success was that the manufacturers of these sensors had to scale these down. So they had to make them smaller and lightweight, and they’ve done a phenomenal job. ‘Cause again, when you had a manned aircraft, weight or size really wasn’t a big issue. But when you’re trying to use one of these small aircraft, particularly weight is a big, big issue. So my hat’s off to these manufacturers. Within a short period of time, they’ve done a phenomenal job. So that’s kind of a summary of those. So, Joe, do you wanna add to that?
Joe Maja: Yeah. So I’d just add that in real-world operation, these categories often overlap. So for example, a system designed for crop monitoring might also support inventory tracking or even mapping for irrigation management. So the value really comes from integration — being able to use one platform for multiple purposes, rather than managing separate systems for each task. So that’s where the technology is heading, and it’s what makes it more practical for growers.
The most-adopted application: simple images and video
Ping Yu: So with all those different types of drones that have been used in commercial nurseries, what would you say is the most used, or adapted, drone that has been used right now in nursery settings?
Jim Robbins: So far and away, I think it’s just for getting simple still images and videos. Because that’s the easiest thing. Or, like I said — we know of a grower in Texas that used to get up and have an employee walk in the gutter of gutter-connected structures, and one of them fell. And so they said, “That’s it. We’re not doing this ever again.” So we’ve demonstrated this at nurseries where, I mean, just a very small drone with a high-resolution camera, and you can do the same thing, but safely.
So those are two low-hanging fruit, that every single operation should have a small drone for those. And it’s not expensive. It’s very affordable for them to do that. But then you start getting into these other things — they’re a little more complicated, ‘cause you’ve gotta invest in people and technology, whether it’s spraying, or whether it’s crop monitoring, or whether it’s this RFID thing that we’re working on. I mean, I think it’s gonna take you a little more effort to jump into those.
Greenhouse vs. nursery — and the case for a practice drone
Ping Yu: We kind of briefly mentioned that most of those works that you guys have done are like in nursery settings, but I know that there are researchers out there — even maybe you guys have done some work in greenhouse settings. ‘Cause I know in greenhouse it can be a little bit challenging, because it has advantages and disadvantages, but one of the disadvantages of working in the greenhouse setting is when you are flying drones — they have a lot of those different structures within the unit, and that can cause some problems. But can you tell us what’s the difference, what are the challenges of using drones in greenhouses versus in nurseries? What are the things that you need to adjust?
Jim Robbins: Yeah. Another thing is just — it’s obvious — the physical infrastructure inside a greenhouse. Some of these can be fairly wide open, but you’re still gonna have posts and columns and suspended things from the infrastructure, that these can be extreme challenges for flying. You also have to think that inside the greenhouse, a lot of these more advanced productions use some kind of a gantry for irrigation or a pesticide application, and it just makes much more sense — if you wanna attach a sensor or something to that same gantry, than to be introducing a small drone. Now, having said that, there are companies that clearly advertise small drone systems for the inside of greenhouses. So it is a reality.
And I will also say that if we flip to the outside, we know of the application of applying, say, shading compound now using a spray drone. Again, instead of getting an employee up on the superstructure to apply these materials, the no-brainer solution, if you’re gonna apply these, is to use a spray drone for applying shading materials. And so that is an accepted norm now in the greenhouse industry. And again, it’s much more efficient, ‘cause you can design the spray pattern precisely, and again, you’re not getting an employee up off the ground. So it’s a much, much safer application.
I should also — just a real quick joined thought here — we talk about drones, but I would encourage your listeners, if they’re gonna get into this and they’ve never flown a drone, go ahead and get what we’d call a practice drone. And they could be less than $50, easily. But I’ve heard this multiple times — people have gone ahead and gotten a Christmas present, a thousand-dollar drone, no experience, they go out, and sure enough, they slam the thing into a house or a tree or something like that. It’s far more digestible to just start out with one of these and get a feel for what’s involved. And so it’s a very worthwhile small expense to just get one of these practice drones.
Ping Yu: Well, yeah — if you have no background in this, it really is essential. People would be happy to get a drone as a Christmas gift. But I was kind of lucky enough to be able to join one of the trips when you guys were doing an on-farm trial with a drone in one of the nurseries in Georgia. So I was able to see how you guys conducted the trial, and the drone in and of itself.
Putting a System Together: Equipment, Payloads, and Bundles
Ping Yu: But can you give us a little bit about what kind of equipment is needed for a drone? I know for different purposes of using the drone, you may attach different pieces into the flying unit. But can you just give us a little bit more about the equipment or the components for a certain purpose of using a drone in a commercial nursery setting?
Jim Robbins: It’s important to point out — let’s say that we follow up with Joe, and we say, “You’re interested in crop monitoring, and you’re gonna use some kind of a three-, four-band multispectral sensor.” Unless you’ve really got the expertise, more than likely you’re gonna pursue a bundle. In other words, there are companies that have already put the drone, the aircraft, together with the sensor. And the reason I point this out is because, again, I appreciate it, because over the years, Joe has done all of this magic for me. But think about this — that sensor would have likely its own power source. You have to have a way to trigger it. I mean, somehow you’d like to have a trigger that’s connected with your remote control, your RC, et cetera.
So there are companies that, if you look, will sell you — it’s already put together — a specific multispectral sensor with a drone, all put together. So some of this electronic configuration work has already been taken care of. I’m not to say that people couldn’t do it, but people forget that you’ve got to have — it’s more than just attaching this payload to the underside. There likely has to be some possible communication with the ground. There has to be a power supply, et cetera. All these things you have to kind of think through. And so, more than likely, again, you’re gonna seek out somebody who’s put this together for you.
Same would be — the extreme example would be a spray drone. I guess in theory, somebody could buy the actual aircraft, buy the actual spray components, put it together, but it’s more likely, 99% certain, that you’re gonna go and somebody’s gonna sell you this already packaged together. But, you know, for just simple pictures, videos, that kind of stuff — and those can be very powerful. Again, we’ve done counting — I talked about the estimating bark-pile volume — that can just be done with a regular RGB digital photograph, et cetera. So you’re talking about a drone with a 4K camera, so somewhere from five hundred to a thousand dollars, something like that. And you’ve got something pretty powerful.
Joe Maja: Absolutely. Integrating sensors onto drones involves more complexity than it might seem. So you have to manage power distribution, synchronize data capture in-flight operation, and ensure that the data is properly stored and transferred. So if any of those components are not properly aligned or configured, the system won’t function as intended. So that’s why pre-integrated bundles are so important. Manufacturers have already addressed these challenges, so instead of trying to build a system from scratch, growers can focus on selecting the right configurations for their needs. So it reduces risk, saves time, and generally leads to more reliable outcomes.
From Image to Answer: Software, Services, and the Economics of It All
Ping Yu: So speaking of the images and videos — those are the things that easily can be adapted by anyone who is interested. But one of the things that I wanna ask you guys is, so for instance, we took the photo, we took the image — and especially for people who are looking into using the drone to do, like, plant monitoring or the inventory — once you take the photo, you have to kind of analyze the data, the image. What are the software that they need to have, to be able to analyze those images and then get those results from whatever they’re taking — the videos or the picture — and get the data that they wanted? Can you walk us through, using one of the examples, either with your, or the plant monitoring example, to kind of give us a walk-through for the imaging analysis?
How the counting workflow used to work
Jim Robbins: Well, I’ll start. So I’ll just specifically use Josue Nahum Leiva’s master’s work on the counting. And so we purchased image-based analysis software called Feature Analyst, and I assume it’s still available. And then there was another one that he looked at called eCognition. And we had to buy a license for that. We also had to buy a sub-license for ArcGIS, I believe it was. But the process was, we would take still photographs, land the aircraft, then just take the micro SD card out of the drone, stick it into your computer, and then download the images. And then, in our case, Josue needed to develop the step-by-step methodology for how to handle and process those images, in this case, to give you a plant count.
Today: upload to the cloud and use a service
Jim Robbins: Now, today, the analogy is — there are a number, and we’ve talked about this, and we could provide names, or you can just search for “plant inventory software.” So there’s a number of companies that offer — what you do is you would do what I described. You would take your digital images, land the aircraft, pull the micro SD card, but you would then upload those images to the cloud. And then you would contact the company and say, “Hey, I’d like you to give me a plant count for this.” And they will do that and send you back a result. So they’re gonna use these software programs to analyze the images for you.
A caution, because we’ve run into this, is — just understand, talk to the company before, because you need to describe. They have tremendous expertise, for example, but have you ever worked with a container nursery? So you’d wanna find out their experience level, because it’s not perfectly transferable cases. The other thing is, they’re likely gonna tell you how many images. So you and I might think, “Well, I’ve got a small nursery, and it’s a small block, and I only need one image.” And I’m gonna predict that the answer is no, that you need, like, eight images. There’s a series that is required. So you’ll describe your situation, and you’ll probably come to an agreement: okay, we need a minimum number of so many images. So all that would be worked out. But to simplify — in theory, you could do this by yourself, buy all the software, learn a technique, but I don’t think most people are capable. So you’d probably, again, wanna use a service to process your image for that specific purpose.
The fully hands-off option
Ping Yu: In that case, if someone wants to do something like that, and take imaging and do the analysis, but they don’t wanna fly, and they don’t wanna go through any of those processes of flying the drone in their operation — are there any services or companies that provide the service, say, “Hey, I want you to come over to my property and then fly the drone and then analyze the data and give me the results”? Are there any services like this, so that for anyone who wants to do that, they don’t have to go through any of the problem, just to give me what I want?
Jim Robbins: Great question. So yeah, we should say that, for anything described, you could easily find somebody to come out and fly the drone for whatever specific purpose you want, whether it’s just to collect digital images. So you could be a totally hands-off process. Call them up, they’ll do it, and then they would coordinate with a different company to process the images for your objectives. Same for spraying. You could just say, “Hey, please come out and spray block C tomorrow at, you know, 3:00 PM,” or something like that. I mean, it’s so easy — if you just search your area, to find drone services for any of the specific tasks that we’ve talked about, you can find them.
A fact sheet on three ownership scenarios
Jim Robbins: And then — this is a great point. Joe and I, we wrote an elegant extension fact sheet. It was probably the final one that I wrote, and we wrote it with an agricultural economist at the University of Arkansas Division of Agriculture, Ron Rainey. And it’s really wonderful, because what we did is we did exactly what you’re describing, Ping. We gave three different scenarios related to a drone, and we used as the example multispectral imaging.
Scenario Number 1 is, you do everything yourself. So you go buy the drone, you get licensed, you buy the software that’s necessary. And it totals it up — how much would you invest in year one, and then it extrapolates it out over a five-year return. And then Scenario Number 3 is exactly what you were asking, Ping, which is a great question: “Hey, we just call somebody up and just come out and do this entirely for us.” And so, you know, we knew actual figures, what somebody would charge for that. And then we had Scenario Number 2, which is a hybrid. So we said, okay — kind of what you were alluding to, Ping — “Hey, I’m comfortable taking the images. I’m not comfortable processing the images.” So we extrapolated that.
So, when we’re done, we could give you a link to that fact sheet that’s still available. And this could easily be applied — even though it’s for multispectral image analysis, it could be applied to a spray scenario. You would just have to change some of the categories that are listed. So the type of drone would be different, but it’s a wonderful template for businesses to use, to help them make a better business decision. What’s really, at this point in time, the best approach for them? So we’ll give you a link to that, but it’s a wonderful example. And again, it brings out the importance of bringing in an economist, because Joe and I both have backgrounds in industry, and we can come up with these crazy academic ideas, but if they’re not functionally gonna be cost-effective, then this is just kind of crazy talk.
Ping Yu: So we always like to talk about the economics. Yeah, that’s one of the things that’s uphill, head on my mind, is: okay, this all sounds great, but if I’m doing it all by hand, how much does it cost? If I’m going to do it on my own, how much does that cost? Because the whole process of running a nursery — there’s a million things coming out every day, and I have to delegate the time and effort to do this. Does it worth it? Or would it be more wise to just hand it off to some expert to do the whole thing for me?
Jim Robbins: And I’ll be quite blunt forward about saying — for example, I am very, very excited long term on the use of spray drones, spreader drones in nursery applications. I really think long term, this is gonna be a game changer. But in 2025, I think if you’re interested in it, the easiest thing is to go find a local company, which shouldn’t be that hard to find, and work with them and kind of explore it at this point. But thinking of the cost of the unit and all of the little details that we’re still working out as far as application rates, nozzles, all of the labeling issues — everything — I would feel more comfortable advising a nursery to work with an outside firm that’s working day-to-day in this. And then as time evolves, I truly believe that these are gonna be very, very significant in chemical application in nurseries long term.
Ping Yu: I have been asked by various growers in Georgia where, at least, they have expressed their interest in switching from the traditional chemical spray to drone application. And they were like, “Oh, should I do this?” But they don’t have a whole lot of available resources to make the decision which way they should go.
Liquid to Granular: Spray Drones, Fire-Ant Bait, and Smart Flight Software
Ping Yu: But speaking of that — can you guys give us, I know from my own experience, for a spray application, you basically have a tank in there. But you also mentioned there are other ways to kind of swap the tank and make the liquid chemical spray drone into a granular kind of spray. Can you tell us, what’s the difference, and can you tell us a little bit more on that? And are there any available resources — I would think there are firms that can do that?
Row-crop spray companies and the case for hiring it out
Jim Robbins: Yeah. I’m just using Arkansas. There are aerial application drone companies now in Arkansas, and shockingly, it’s in row crop agriculture. And they’re so busy that my colleagues — again, I’m retired — that are trying to work with these companies to work on specialty crops. So for example, I just saw a video from applying a fungicide to pumpkin production. And think about this: a vining crop, so whether it was squash or pumpkins or something like that, there’s no rows. It’s just a field full of sprawling vines. And so if you needed to apply something [laughs] over the top of these things — so what they’re doing, these for-fee spray aerial application companies, is to come in and make the application, and then they’ve got the spray cards, and they’re doing the data collection. And so even as a researcher, I think that’s the smarter way. The regulatory hoops you’ve gotta go through —
By the way, this example — this will be the last thing I’m gonna say about it — I couldn’t believe it, but it’s got an 18-gallon liquid tank, which, these things are — because that’s a lot of weight. And that’s definitely not considered in the small category. But again, in the nursery industry in the south and southeast, application of particularly fire ant bait materials — boy, man, that’s — to me, I’d give that a — because you can define the flight path so precisely, and because it’s a lower bulk density, you can carry a lot more. I just think it’s a really great way. And again, credit to the manufacturers, as I stated earlier. It’s amazing how they just immediately realized that having a very pull-out, transferable unit from a liquid to a spray application system — I think they’ve smoothly made that interface. So, Joe, you got any thoughts?
The engineering behind the swap
Joe Maja: Yes, Ping. So from an engineering standpoint, the ability to switch between liquid and granular systems comes down to standardized interfaces, both mechanical and electrical. So what’s improving now is how efficiently the systems communicate and how precisely they can control the application rate. So there’s still some room for improvement, especially in calibration and — and Ping — consistency. But I believe the progress has been very significant.
Ping Yu: So basically, whether they want to use it to spray a liquid chemical, or a granular, or a weed control product, or anything related — there are products that are available for the growers to choose from, right?
Jim Robbins: Yeah. There are platforms that are readily available. And again, let’s say all you want is — they’ll just sell you that, and then later you could buy the liquid application system, which would be a tank, and obviously it would have a traditional boom with nozzles, so the configuration varies. So basically, you can buy different kinds of components separately.
Ping Yu: And I guess what I’m saying that you have to think is, you just make sure all those components are compatible with your drones, right? Is this something — is this an issue with some of those products out there?
Jim Robbins: No. I mean, obviously, as long as it’s their system, they’ve designed those payloads — the liquid application system or the granular — to work. Again, remember, this circles back to what I said. Think about this: they have to give thought about, how am I gonna power both the sprayer or the spreader? ‘Cause it’s gonna require some kind of battery power, and they can either pull it from the aircraft itself, or they can provide, most likely, a source. And then you also have to have a trigger — actually want this system to turn on and to turn off. And so that is all being taken care of for you, thanks to great engineering.
Flight navigation software: the biggest advance of the last decade
Jim Robbins: And the other beauty about these new systems — it’s absolutely remarkable — is, again, the flight navigation software. To me, that’s been the most significant advance in the last ten years. So now they’ve got sprayers that will lift off, go apply. They’ll sense that they’re getting low on material. They’ll stop spraying. They’ll come back to the starting point where you can refill. They will then go back to where they stopped and continue along the path. So this is kind of seamless. And the same is true for photographs. You can see in advance — okay, if I need to capture images of this area, it’s gonna take maybe, say, 28 pictures. What’s the length of the flight path? Do I have enough battery? All of this is now seamlessly thought through for you. So, again, circling back, 15 years ago, oh my gosh, it was really rudimentary, crude. And again, Joe did all this stuff in the earliest years. So credit goes to Joe and his many years of engineering.
The Ornamental Challenge: Diversity, Variability, and a Software-Smart Future
Ping Yu: So speaking of that — and I know Joe kind of was dragged into horticulture by James — but I know right now, a lot of times, for instance, for drone application in row crop, most of the time it’s monoculture. It’ll be easier for them to kind of build up the kit and then just fly and apply whatever they are applying. But in horticulture in general, we’re dealing with thousands of different types of crops, and they have different requirements, and some of them might be very sensitive to certain of the chemicals that you’re applying. Would you say this would be one of the challenges, or are there any solutions to address the challenges and make it a little bit more adapted by the grower? For any manufacturers out there who wants to do more business with ornamental horticulture, what do you think was your experience? What are the things that you think need to be addressed in that perspective?
Jim Robbins: I’m gonna jump in here real quick, and then let Joe — but just think about this. If you’re growing 3,000 acres of soybeans, or 3,000 acres of cotton, a monoculture like that — it’s a heck of a lot easier than if you go to a nursery that’s growing 3-, 4-, 5-, 600 different types of crops. You know, Joe and I have done work in several different ornamental crops in containers. And I think it’s — what we learned on panicle hydrangea — well, that’s the challenge against us working on this multispectral area. We have a handicap, in that we’ve got a much larger diversity of crops than someone in row crop, where it’s thousands of acres of one thing. And so some of their solutions are gonna be achievable sooner than they will be in our ornamentals area.
And, you know, in a rough sense, maybe we’d be able to find, with a thermal sensor, that there’s some general water stress before it occurs, or something, on a diversity of crops. And I think artificial intelligence will help us long term. But yeah — again, there are so many more cheap ornamentals compared to other situations. What are your thoughts, Joe?
Joe Maja: So in ornamental systems, the challenge comes from, I believe, variability. So each species has its own spectral signature, influenced by factors like leaf structure, pigmentation, and canopy architecture. So a model that works well for one species, or what’s called a taxa, may not perform well for another taxa. That makes it much more difficult to apply a single solution across diverse crops, and it requires more advanced data processing and modeling approaches.
Ping Yu: Well, I guess the ornamental, with the diversity of plants that we’re growing — it’s a beauty, but sometimes in this scenario it’s a challenge, too. But do you think, with the involvement of the technology down the road — let’s say in the next 10 years or 15 years — are there any possibilities of solutions that can help make it possible, kind of reduce, or solve the challenges we are facing now in terms of the diversity of crops? And for instance, do you think it’s possible for people to come up with ideas to use, like, different types of drones in, like, say paniculata, roses, and then Japanese maple, and then go from there? ‘Cause that can be maybe a little bit more expensive than monoculturing row crops, but do you think it’s possible down the road?
Joe Maja: I think the future solution is less about different hardware and more about smarter software. So AI and machine learning will play a key role in adapting to variability across species. So we’re also seeing advancement in hyperspectral sensing and onboard processing, which actually allow more real-time analysis. So now we have the so-called edge computing. So the hardware may remain relatively consistent, but the intelligence behind it will continue to evolve.
Closing
Ping Yu: Conversation like this only happen when you support the show. For more information and to find ways to support us, please go to bandbpod.com — bandbpod.com. If you like what you are hearing, spread the word and share this podcast with the people you love. Make sure you hit that subscription button and give a review for the podcast. And as always, go check out the show notes to learn more about this topic and other topics we featured on the show, at bandbpod.com. Thank you for listening. Till the next time, stay healthy and go plants.
Transcript produced for Blooms and Beyond / UGA Center for Urban Agriculture