Co-Founder & CEO of Progeny Drone
This time around our spotlight has turned towards Anthony Hearst.
We would love to thank him for agreeing to take part in this interview amid busy schedule.
He is the Co-Founder & CEO of Progeny Drone, that has launched a software app that allows agronomists, plant breeders, and others concerned in small plot trials to easily and quickly convert aerial crop pictures into live plot-level metrics of plant health and growth which can help researchers take more data-driven choices regarding best practices.
In this interview, Anthony will talk about how he got into drones, what inspired him to launch Progeny Drone, the future of drone agriculture, a to-do list for his application, advises for entities looking to incorporate drones for agriculture applications, and so on.
So, what are you waiting? Without further delay let us dive in!
Tell us a little about yourself. How did you first get involved in drones?
I’m a recent graduate from the Purdue University Department of Biological Engineering & Agricultural (Ph.D.).
I first became interested in drones in 2012 during my undergraduate in Chemical Engineering at the University of California, Santa Barbara, in which I worked as a research assistant for a lab focused on satellite-based remote-sensing.
I was conducting a literature review and read a few of the first publications on using drones to map outdoor environments with unprecedented levels of detail. It sounded like an exciting new opportunity for research. I was lucky enough to be accepted into Purdues graduate program when they got their first drones in 2013.
What inspired you to start Progeny Drone, initially?
During my first couple of years working with drones at Purdue, we found it extremely expensive & hard to convert our raw drone images into accurate & useful plot-level metrics of plant growth & health. It soon became apparent that even research groups with a great deal more funding and personnel were struggling just as much as we were.
This was because of the large amount of data involved, the difficulty of stitching images of crop fields, and the many different ways in which field trials get planted, labeled, & analyzed (e.g. 1-row plots, 2-row plots, 4-row plots, 8-row plots, marked by row & range number, with measurements performed on certain portions of specific rows within each plot, etc.).
From around 2014 onward, I knew someone would need to develop practical & affordable software for this specific but essential application of drones, and I had some ideas about how this could get done.
But, it wasnt until 2017 when my co-founder, Dr. Katy Rainey, associate professor of Agronomy at Purdue, suggested I apply to Purdues Ag-Celerator competition, that I ever thought I might be the person to develop & distribute such a software tool.
Getting accepted into the Ag-Celerator competition and going through their 6-week FireStarter business course helped shake me out of my academic mindset and become interested in the startup world as a means of helping people and making an impact on the planet.
From the moment we won $60,000 from the Ag-Celerator, I was committed to this career path.
Tell us about the software application you developed for plant breeders? How does it work and what is unique about it?
Progeny was designed to be a fast, simple, & low-cost solution for agricultural researchers conducting outdoor small plot trials. It includes plant breeders, weed scientists, seed treatment researchers, soil treatment researchers, crop consultants, contract researchers, & drone service providers.
Progeny does this by eliminating the need for high-grade (centimeter-level accurate) GPS, cloud computing, or internet access from the processing workflow.
Instead, researchers with no programming experience can use low-cost drones (around $1000) with low-grade (10 meter-level accurate) GPS and a laptop (under $1500) to convert high-resolution imagery of their field trials into custom-labeled plot images and metrics in 10 minutes while in the field.
Specific output metrics we currently offer are canopy cover, greenness, stand counts, & row-length.
All the researcher has to do is supply progeny with some necessary information about the layout of their field experiments and click the corners of their tests in a few images.
This new and simpler workflow is possible thanks to patented image stitching, plot extraction, & analysis algorithms we developed at Purdue University specifically for this application.
Tell us about the team behind this successful software application?
Since Progeny is currently a small startup with limited resources, we decided to use our funding from the Ag-Celerator competition to hire professional contract developers from another Purdue-based startup, Delmar Software Development LLC, to convert the source code we developed during our research at Purdue into a user-friendly standalone software application.
We set up so I can quickly prototype new algorithms in a programming language I am comfortable with, while the developers take care of integrating it into our app and maintaining the license server.
It allows Progeny to benefit from our deep understanding of the technology & years of experience collecting & processing drone imagery, as well as the years of expertise Delmar has with professional software development and deployment.
It allows me to quickly respond to customer requests and make sure everything is working seamlessly and smoothly from the start. Ultimately, our success comes from the quality of our customer service.
You’ve earned a doctoral degree in biological engineering and agricultural at Purdue. What exactly is drone agriculture, and can we expect to see more of that on the rise?
There are currently two very different types of drone uses in agriculture. One is large-scale crop monitoring with high spatial coverage (thousands of acres), but low spatial resolution (10 cm/pixel), for farmers. The other is smaller-scale (tens of acres), but higher-resolution (1 cm/pixel) crop monitoring for agricultural researchers.
For now, Progeny is focused on adding to the list of plot-level metrics of plant growth and health that agricultural researchers can quickly and easily get from high-resolution images using our app.
For example, we will soon be offering plant height and weed/crop cover in addition to canopy cover, greenness, stand counts, and row-length.
Our long-term goal is to enhance our algorithms to the point that they can be used by farmers to make real-time management decisions. We believe that farmers will adopt the technology after researchers demonstrate its practical value.
What are some of the typical backgrounds of your clients? Do you get a lot of plant breeders, or is it people looking to conduct small-plot trials?
Currently, our clients primarily are plant breeders & contract researchers who have already purchased low-cost drones but have not yet been capable of getting much value out of them. We help them start getting a return on their investments in drone technology.
Agriculture is among the fastest-growing sectors for drone users. First, why are drones getting used so extensively on the agricultural field? And are there any certifications or individual licensing a UAV operator requires to fly drones to agriculture?
I think there are two main reasons why drones are being used heavily in agriculture. First is that precisely measuring plant growth & health, particularly at the plot-level, is slow, labor-intensive, & subject to human error.
It is because research trials are regularly planted all across the world, and researchers must visit tens of thousands of plots about 1-5 times each year and subjectively score them on a 1-5 scale.
Drones can increase the measurement precision by at least a factor of 10. They can also increase the measurement frequency, with many researchers flying drones over their field trials 5-15 times a season.
The second reason is that agriculture is an industry where both researchers and farmers are regularly making risky decisions based on very minimal, low-quality data.
The more data on plant growth & health researchers and farmers can get without too much difficulty & expense, the better they can manage their operations, reduce their risk, and increase their profits.
Why do you feel existing small-plot trials techniques are ineffective? How your application able to address these issues and speed up the process both in terms of time and money?
Measuring plant growth & health by-eye is expensive and slow because it requires a trained agronomist (often with a Ph.D.) along with a team of field assistants to travel to every field site and walk through and carefully look at tens of thousands of individual plots.
It can take a full 8 hours to collect plot-level measurements for a single field site, and the quality of the measures often drops as the day drags on.
With Progeny, a licensed pilot with a $1000 drone, a $1500 laptop, and only a basic understanding of the layout of the field experiment can collect the same metrics ten times faster with ten times greater precision.
By adopting this new technology, research groups can quickly save thousands of dollars just in operational expenses, let alone the benefits of having a faster turnaround of more precise metrics.
What features or technological capabilities differentiate agriculture drones from consumer drones?
Agricultural drones differ from consumer drones in that they typically have higher-quality cameras and are intended for capturing hundreds to thousands of high-resolution, overlapping, geo-tagged images of crop fields from nadir-view & constant altitude that can be stitched and converted into precise metrics of plant growth & health, rather than just taking a few low-quality snapshots.
You mentioned Progeny will turnaround data within 10 minutes. What’s the mechanics & specs behind your application that leads to this much quicker processing compared to traditional alternatives?
Part of the reason why Progeny can provide such a fast turnaround is that it takes an entirely different approach to extract individual research plots from the raw frame photos.
Instead of taking the raw images, combining them into a single giant ortho-mosaic of the entire field site, and then overlaying a shapefile delineating the research plots in the mosaic.
Progeny can skip the step of generating an ortho-mosaic and instead extracts ortho-images of individual plots directly from the raw frame photos in a manner that is highly parallelized.
It is much less memory-intensive and also avoids much of the blurriness or geometric distortion you often see in ortho-mosaics of crop fields.
What is in your to-do list for the application? Being a startup, are you getting the necessary funding to keep improvising the platform? How people could help you out?
My current to-do list for the application is adding plant height, and weed/crop cover to the list of output metrics researchers can get through the app.
After that, I will work on adding a time-series analysis module so researchers can start analyzing plant properties over time. Then I will work on adding flowering and maturity date detection.
Right now, more than anything else, we need help finding customers. The money we receive from customers will enable us to continue developing the app to meet customer needs.
We know there are a lot of small plot researchers out there, but they arent yet aware that we exist. Once we get a chance to show researchers what Progeny can do for them, that is usually all it needs to get them going on a free trial and eventually purchase a license.
What is the market for drones and agriculture right now? How has it changed since you started your business?
The market for drones and agriculture is currently mainly focused on large-scale crop monitoring with low-resolution images.
There are a few drone service providers who will extract plot-level metrics from high-resolution photos for researchers, but they charge tens of thousands of dollars.
There is currently no app aside from Progeny that enables researchers to quickly and easily extract plot-level metrics from high-resolution images themselves at a small fraction of that price.
One way in which the market has changed is that now both farmers and researchers are much more wary of drone products as they have been let down by a lot of the current software and service providers.
Fortunately, I believe both drone hardware and software are finally past the point of being prototypes and can now be affordable, reliable, and useful tools when used correctly.
The challenge now is showing researchers that we have something that can save them a lot of time and money. We need a few more customers to start gaining traction.
The annual meeting of the National Association of Independent Crop Consultants in Savannah, Georgia was the most exciting event I attended last years with regards to being a place where many contract researchers who regularly conduct outdoor small plot trials gather and are willing to look at new technologies. Phenome in Tucson, Arizona, was also a good one for finding plant breeders.
This year, if I get a few more customers, I hope to travel to other conferences across the world where I can meet different types of researchers such as weed scientists, seed/soil treatment researchers.
For a person or company who is considering a drone for agriculture applications, what suggestions do you have for finding the right product?
For research, I highly recommend a low-cost rotor-copter type drone (around $1000) with a high-resolution visible (RGB) camera like the 20MP DJI Xenmuse X4S.
Rotor-copters can fly lower than fixed wing platforms and capture sharper images. They can also avoid unnecessarily encroaching on neighboring fields which may make it easier to secure permission to fly.
For a computer, I recommend a Windows gaming laptop under $1500 with 4-6 cores and a solid-state drive (SSD) for portability & speed.
I would advise against purchasing anything more expensive than this until you have had a chance to learn what is possible with this equipment and Progeny.
In the future, how much of an impact do you think agriculture drones will have on our everyday lives?
I think drones will have a tremendous impact on our lives for recreation, data collection, and even transportation and delivery of specific goods, but not until air traffic control technology & regulations have gone through some significant changes to make them safer & more reliable.
I dont think this will happen for anything outside agriculture and construction for at least another 20 years. But, hopefully, Im wrong about this.
For those people or companies who are using agriculture drones in their business, what actions should they take to safeguard themselves against liability?
Get licensed and registered with the FAA here. Never fly in restricted airspace. Get a low-cost but widely used and recommended model of rotor-copter with a decent RGB camera (no expensive multispectral or hyperspectral cameras or Lidar) and fly it low.
Do not try to fly when wind speeds are too high for your platform. Regularly inspect your drone for signs of wear/tear and replace propellers early/often. Take good care of your batteries.
Bring a laptop and use a free trial of Progeny to process your data in the field, so you know you got what you came for before you leave a field site to avoid having to revisit/re-fly field sites.
If you want to provide flight services for researchers doing small plot trials, please get in touch with Progeny, and we can help you understand what they are looking for and how to collect the imagery so they can get good results with Progeny.
For a business that is just beginning to incorporate drones to their agricultural land, how may they proceed about things? How may they find out the best product for their requirements? Do they want any software like yours to make the best usage of UAVs?
If it’s a business that conducts a lot of outdoor small plot research trials and has already purchased a drone, then I recommend getting in touch with Progeny for advice as well as a demonstration of what Progeny can do with their images and a free trial license.
What advice can you offer new drone operators who are entering agricultural and biological fields?
The best thing about Progeny is that it works with any low-cost drone camera. It creates an opportunity for pilots willing to fly over field experiments so researchers dont have to purchase their drones or travel to field sites as often, which could save them a lot of money in travel expenses among other things.
I can help pilots take advantage of this opportunity by connecting them with researchers they may not know about who are in great need of and willing to pay for frequent flight services during the growing season.
These researchers may be ready to purchase a license for Progeny, which they could temporarily let the pilot use on their laptop for in-field processing & data quality control before delivering images to the researcher.
Alternatively, pilots may consider purchasing a license for Progeny and using it to provide plot-level images and metrics to multiple agricultural researchers, but this is not recommended until the pilot can find enough stable clients to make it worthwhile.
I am also willing to grant free licenses to pilot training programs to help pilots learn more about this opportunity and develop the skills needed to take advantage of it.
What qualities do you look for in a drone when integrating with your software? How do you measure the performance?
The ideal drone would be able to fly low & steady, even under windy conditions, and capture nadir-view geo-tagged images that are as high-resolution and well-focused/sharp as possible.
The less lens distortion is, the better. Also, the more control the drone operator has over the internal camera settings, the better, as the default settings on most cameras are not the best for precisely measuring plant growth & health.
Unfortunately, most drone companies currently do not allow their users access to their internal camera settings.
You mentioned that you interviewed hundreds of users from weed scientists to plant breeders. What did you first notice about your product that made you think it was going to catch on?
The first thing I noticed was it would work with the drones/cameras they often already had and would let them collect plot-level metrics without having to do any field-site preparation or change the way they have planted their experiments. It is easily compatible with their existing equipment/workflows.