Menu Close

Students Build AI-Powered Drone for Earth Observation | NASA SEES Internship

Video Summary

In this interview, four high school students from the 2025 NASA SEES (STEM Enhancement in Earth Science) internship program discuss EarthLens, an innovative drone-based system they developed to complement satellite Earth observations with high-resolution ground data. Samuel Bawden, Nandini Khaneja, Neev Tamboli, and Jordan Rodriguez share their journey from classroom learners to hands-on innovators, demonstrating the power of experiential STEM education in workforce development.

Project Origins and Technical Design

EarthLens emerged from the students’ experience using GLOBE Observer to compare ground-truth land cover data with satellite datasets from Landsat and Sentinel. Recognizing the challenges of manual data collection and the vast difference in spatial scales between ground and satellite observations, the team designed a drone platform that bridges this gap. Their system integrates STELLA-Q2 and Helio-STELLA spectrometers for measuring surface and atmospheric properties, RGB cameras, a Raspberry Pi-based data system, and a preliminary AI classification tool for real-time land cover analysis. The goal is to make Earth science data collection more accessible for citizen scientists, educators, researchers, and agricultural applications.

Learning Through Hands-On Experience

The students candidly discuss their learning journey, emphasizing that none of them began with significant remote sensing or drone-building experience. Through hands-on problem-solving, mentorship from University of Toledo graduate students, and self-directed learning, they assembled the drone from accessible components, integrated unfamiliar technologies like Raspberry Pi systems and flight controllers, and developed software solutions. Their discussion of comparing satellite and ground observations provides authentic insight into how students build conceptual understanding of remote sensing principles through real-world experience. The team emphasizes that their work complements satellite missions, filling an important niche in spatial resolution and data validation.

Future Directions and Impact

Looking forward, the students plan to create build guides for others to replicate their system, integrate additional environmental sensors, and refine both hardware and software components. They envision applications in agricultural monitoring, environmental research, and education, where the system could provide detailed observations at scales between satellites and ground measurements. Each team member reflects on how the project influenced their career goals in engineering and Earth science, demonstrating that high school students can develop meaningful tools for real-world environmental challenges. Presented at the American Geophysical Union conference, EarthLens exemplifies how project-based learning and mentorship cultivate the next generation of Earth science professionals while students continue building their foundational knowledge in remote sensing and related fields.

Transcript

I’m here with the, NASA SEES group that I met at AGU. Working on the EarthLens project.
And I’ll save that for you to tell us all about what EarthLens, is and what you did and all that.
But, with, that being said, I’m going to ask you to, introduce yourself.
Where you’re from again, just the city is fine. And your role in the EarthLens project.
Okay. And I’m just going to go in order from how I see it in my screen over here, from right to left.
Okay. And so first up, I have Sam. Yeah. Hi.
I’m Sam Bodman from Washington, DC area.
I’m, going to Dartmouth next year. I’m going to be a freshman there. And my main responsibilities
with the Within the EarthLens team was, essentially doing,
a lot of the background research for essentially identifying the problem we were seeking to solve, which was improving,
citizen science tools for land cover. So we were interns, in a program this past summer,
which I’m sure we’ll talk about. And a lot of the tools we used were challenging and give great data.
And so kind of assessing what those challenges were and then how to solve them with EarthLens was kind of
where I came in then next up I have Neeve. Yeah. So I’m Neeve, from the Bay Area in California,
about like 20 minutes out of San Francisco. And, my main role within the EarthLens project was
sort of, helping design and plan out the drone as well as designing
how our entire system would kind of be integrated with the computational part. next up, I have Nandini.
Hi, I’m Nandini, I’m from Long Island, New York. My role on the EarthLens team was, was, hardware lead.
So I did the physical construction of the drone and the STELLA integration. And we’ll be continuing to work on various iterations of the drone.
And last but not least, Jordan. Hi. My name is Jordan Rodriguez, and I’m from Westlake, Texas.
And, on the EarthLens project, I focused on software and data architecture. My role involves designing how observation data,
is captured, from the drone and transmitted to the actual app.
first off, I want to understand what NASA SEES is and how, you know, folks from all across the country,
you got to get together and work on the project. So can you tell me just a little bit more about NASA’s SEES
and, and how that happened? Sure. I love to take that question. So NASA SEES
is, really was a gift of a program. I think for a lot of us, they do,
all kinds of programing for high school, students interested in engineering and earth and space sciences
and physics and, so students from all over the country, every corner.
You know, this group is representative of that, but not in the least, limited to how many places people came from,
including, you know, territories like Puerto Rico and things like that. People from all over the country,
got together and worked on a lot of really cool projects. Or specifically was about,
essentially assessing the agreement between ground truth land cover data and satellite derived land cover data.
But there were groups who worked on designing future missions to moons of Jupiter
and Saturn or other really, you know, just crazy cool things like that. And so I think it was a really awesome way to get introduced,
to, you know, kind of what careers in Stem look like. And I think I speak for everybody when I say
it really encouraged us to pursue this kind of work in the future.
Awesome. Awesome. That’s fantastic. And, so we’ve touched on a bit of what EarthLens is, but,
could I bother, one of you to, tell me exactly what EarthLens is? What did you do for the project?
You know, sort of just, the in broad, definition, a broad scope of what the project is.
Yeah, I could take this one. So, I think from a very basic quick perspective, EarthLens,
we came across the idea once we realized how, a lot of times, inaccurate
and inefficient data collection of land cover was with the existing technology
because it relied mostly on, like, manual or just simply very far off inaccurate satellite satellite data.
And we wanted to change this by making it a more accurate and be more reliable, because one of our friends, Jordan, over here, I think he lost
over 200 images that he took over like ten plus hours. And we also had the data that a lot of the pictures
that we had taken, as part of our internship, they were I believe, less than one megapixel in resolution.
So we wanted to see how we could improve that. And we came up with the idea that a drone could address
both of these problems, because a, a drone is kind of like a happy medium between the very inaccurate high up satellites
and the having to go on your feet with your phone manual like, NASA GLOBE.
And with our computational parts, we can also make sure that the data can be stored very safely, and we’re not going to be losing it.
Okay. And so you said inaccurate, satellites. What do you mean by that?
So I think very briefly, one of as Sam mentioned, our internships goals
was to see the difference between, the land cover
that we can see on the ground and what satellites are, showing them to be on the main NASA data sets.
And I think we can probably pull this up later sometime, but we had these large charts that we made that compared these different, strips of data.
And we were able to see that the, the data that the satellite gave about the land,
specifically stuff like spectral resolutions and what was exactly being covered by that land was often inaccurate.
And it would be inaccurate in the way that, let’s say, it would classify something that’s grass as trees that were nearby wouldn’t be accurate
enough to see that, hey, this is a patch of grass over here in the midst of a forest.
Sometimes even, like, you know, shadows that would be picked up, it just wouldn’t give any information.
Or maybe it might even say was like water because of the darkness. It compared with other, water bodies to those, shadows.
Yeah.
tell me how this whole process, sort of came together and how you all started working and collaborating together.
Sure. So the NASA SEES internship program as a whole is open to high school juniors and seniors from across the country.
So we had had the ability to select teams we were interested in, and we ended up getting sorted into those as we got accepted into the program.
So we were part of a 50 person team called Earth System Explorers that met virtually over two months. June to July 2025.
We had a variety of mentors, including rusty, Peter and all the scientists who, you know, and since our main goal was to compare land cover data that we recorded
using the GLOBE, the GLOBE Observer application in our own community, and comparing that to data existing in ArcGIS
and manual, data that we’ve classified in Collect Earth Online.
We, a lot of groups were deciding what to do with that data that we had generated throughout the internship,
since we had like 37 different points within a three kilometer grid in our own communities. So a lot of people wanted to analyze the data that they had generated.
But we took a closer look at our, data quality itself. And then once we saw what a big issue that data quality was and how Jordan lost
such a significant number of the photos that he had taken, we realized that GLOBE Observer had been a very valuable, tool for the time that we were using it.
But it had, you know, there were challenges that were beginning to surface, especially since it hadn’t been
updated as frequently. And we thought we wanted to combine the benefit of the bird’s eye view that a satellite gives with the increased granularity
you would get just going by, manually taking the photos. So we wanted to, take a more unconventional hardware
based approach to doing this. And it was definitely a challenge because we’re all across the country.
So, it was a group decision to go all in on this one leading up to AGU.
what interested you all about combining drones, remote sensing and citizen science all together into one package?
the technologies that have become accessible to, you know, the average person have, improved greatly.
I think, you know, these STELLA sensors and the ability to build a drone in your garage and access to AI tools is something that’s new.
And so why not take a new approach to, land observation, Earth observation.
Yeah, that hits the nail on the head. So for me, I just, we remember how initially we,
I think when we were working with our working group to see what we could take the project further with.
We definitely wanted to keep citizen science in mind, because that was kind of the core value that was ingrained with NASA GLOBE
and, you know, working after the NASA SEES project where we were so deeply engrained with land cover
and wanting to revolutionize that sort of sector, we decided that we had like a few different options.
And I remember Jordan had this great computational idea. And Nandini brought to, a hardware based idea where we had,
like, instruments like STELLA and then we looked at each other and we were like, wait a minute. We can kind of combine both of these to make like an intelligent hardware
based system that uses this type of instrumentation to solve the problems that we need to.
what was your experience with, like, say, drone technology, remote sensing and or the, STELLA instruments themselves?
Sure. I’d like to talk about that from from, like, a hardware perspective. I had never built a, never built a drone before.
I actually, my, mechanical engineering related experience, even before the internship program, I had done a research project,
working on a martian simulation chamber in my basement with a friend of mine. And that, you know, made me very familiar with that low cost optimizing,
being responsible for basically every type of engineering in one project kind of experience.
So from there, I, we were able to, I researched how, I guess these kinds of drones have been built in the past.
I visited my local hobby store to learn how to mount the motors. Right. And, you know, try to use online resources and my own community
to try to make this hardware as accurate as possible. And for example, we were we ourselves were not familiar with STELLA tools beforehand.
And I personally also had never used a spectrometer board. Actually, we started using STELLA because it gave us a place to start
exploring this. The first step, the very first step we took hardware wise, was ordering every part
that is required for the Helio-STELLA and the STELLA-Q2 designs. Attaching the circuitry, making sure it worked
on the SparkFun boards that it’s designed for. And then once I realized we needed to integrate that
into a Raspberry Pi system to record that microSD card and be part of the transferring to Jordan’s app system,
we were able to slowly strip away components like the real time clock didn’t need to be needed anymore.
The, of course it’s slipping my mind right now, but there’s displays, there’s much, many components we were able to eliminate
to make it to a more barebones style model that hopefully any, lab worker or hobbyist could build in their basement for a,
reasonable sum of money. Yeah, I think the cool part of it is none of us came in
with any significant experience directly related to this. And I think in, in a, in a cool way,
that’s an advantage on the citizen science end of things. Because if we can make something that is intuitive,
then it can be widely used. Yeah. And just focusing on like even though we were so inexperienced
with hardware like this, I think one of the beauties of science and in specific here with STELLA was that there’s like this community
around people that know what they’re doing. And for example, we were able to reach out to grad students all the way out in Ohio, and they were able to help us
because they were like STELLA experts, and we were able to use their help. And that kind of brought us forward to a stage where we wouldn’t
have been able to without them. Oh yeah. And of course, of course. It all ties back to you know, the, the SEES program giving us
this opportunity to explore what we probably wouldn’t have otherwise known,
if it wasn’t actually for the SEES program.
Rusty connected us, with two graduate students at the University of Toledo, who go by Femi and Feisal,
who, they work under a professor that she knew very closely. And they were both very helpful in two different ways.
So for example, Faisal was able to help verify, our technical planning,
while Femi was very helpful in making sure that we knew the, just generally how to plan a project like this, because it’s
such a big undertaking and especially considering we’re for high schoolers, trying to figure it out because, I mean, in a way,
we are the citizen scientists trying to put together this kind of a system. So I think having so much support on that side from,
not just the scientists that we directly worked with in the SEES program, but scientists, we were able to be connected to, during the design process and even after.
were there any other, like, major components of the EarthLens systems that you developed, that you didn’t previously mentioned?
I think, you know, the we did a lot of work,
specifically with how to fly the drone. With Nandini, you can talk more about it, but essentially,
you know, you want it to be, clean user interface, and you don’t want it to be, you know, clunky controls
where you have to type an input for the for the drone drone to go up or down. And so there was a lot of tinkering we had to do on that.
And Nandini you want to talk about that, that’s cool. And it one that feeds right into the next question, which would be like and I figured to go right into there,
what was the sort of, most challenging technical problem you had to solve? And how did you approach it?
So I got to say, it’s like, I think it’s the combination of all the little problems that came in along the way.
Like, for example, we’re using this Raspberry Pi based system. We haven’t used Raspberry Pi’s in the past, past all my experiences
with Arduinos. So I think understanding the Raspberry Pi itself was going to be like a mini computer and required its own monitor
and keyboard, and that kind of stuff was not something I expected. So we ended up working around that so I could run it from
my own computer connected to a hotspot on my phone. So that it can be done literally anywhere, because this was because
we had to do early stages before it gets fully connected to just an app. So like that’s one of the challenges that came in the way our flight controller,
for example, they come in with preloaded software sometimes. And, you know, these things are usually designed
for like the first person view racing drones. And it was definitely hard to see how big of a how big a propeller,
what kind of flight navigation hardware, what kind of, GPS system is required for an actual observation drone.
So we had to take risks like deleting the software already on that, on that flight controller,
like $140 flight controller, and hoping that I downloaded the, the GPS navigation appropriate software.
INAV correctly instead. So I think it was just the combination of, you know, we can very clearly plan
this out and go, okay, step A, step B you know, these are going to be all the steps that we’re going to take.
But there are so many little not necessarily detours but little bumps in the way that I guess
you become more familiar with once you or a small group is responsible for the entirety of a project like that at this scale.
So I think that was definitely the biggest thing to get used to.
So yeah. And you were integrating the STELLA-Q2 and the Helio-STELLA spectrometers into your platform.
And you also hinted of this, but, what did you learn from that process and why did you do it?
So the reason I think we chose those two in particular was that we we noticed that the GLOBE Observer app was not just land cover.
It also had a cloud cover function too, even though we ourselves had used it. So by we already planned to have a camera facing down and a camera facing up,
which that’s kind of being worked on still, with the multiplexer,
but so we wanted something that was going to measure the upward and downward, both because you wanted to have something, you know, comparing it to.
So it’s not just one data source. So like for example, like I have the actual thing here too.
We designed it using the same cables in the STELLA system. I don’t know if the blur will let you see it. So that we have the triad spectrometer used in the Q2 going down
and giving us more channels and more and more information or data like we would need for that land cover observation.
While we would have, I believe, 8 or 9 channels something, something a smaller number of channels
in the Helio-STELLA one and those two combined. Well, because we were able to do that,
like the daisy chaining of storage with the quick cables. So I through the STELLA system, I learned how to even do that,
because if I just looked at the spectrometer board alone before that, I would be asking myself, where’s the ground?
You know, where’s the ground? Where’s the power? Just working and very like, simple. Like the normal DuPont wire setup.
So I think, especially that STELLA the the guides for STELA online
help me see how intuitive the connection, especially between these two specific ones, would be and how valuable they’d be for implementation.
Yeah, and it’s not just how intuitive the actual hardware is, but also just the data that it spits out.
So like for our initial tests, like using the spectrometers and photos,
we were literally able to create charts of our of our first data collections by just copy and pasting values
from the output into a Google spreadsheet, Google, Google Sheets spreadsheet. So it’s not so, so STELLA.
The STELLA devices, you know, they served our purpose really well of wanting to get, you know, all of the data about light and air
and not just the ground and not just the sky, but how the lighting, essentially mixes with the irradiance between those,
two sources, interact with each other. But also just how easy it is to get the data from the STELLA hardware
to, a format where we can interpret it.
so you used basically the Helio-STELLA to help normalize the data you were getting from the surface reflectance and all that.
Okay. So sort of instantaneous reflectance. So you just bypassed, the whole irradiance problem itself.
I’m sure you have quite a bit of uncertainty with those, because those are not the most sophisticated sensors ever.
Are you taking a look into those, into the uncertainties as well? So, yeah, there there was a certain amount
of normalization we had to do, for example, the, the, the, the wavelengths that each, of light
that each of the two spectrometers collect are not exactly the same.
They’re only off by a little bit, which means in a graph, you know, we can we can, pretty accurately compare them.
But there is that small contingency you have to deal with. And then, you know, like you said, it’s we’re not working with,
you know, massive, you know, super accurate spectrometers. So it really is the other thing that we’re working on with
the drone is how we can collect the maximum amount of metadata, how high we are, how fast or moving
temperature, time date and all of those things
combined to kind of account for anything that might not make sense. And so you get a much better picture by using the other aspect,
the other pieces of data that the drone itself collects to get a better picture of what the STELLA spectrometers are seeing as well.
Yeah. And one of our like, further works that we’ve planned out as well is kind of integrating an AI model into our Sis system.
And what this could do is make use of all of that metadata, as well as the raw spectral data
that we have to kind of eliminate as much as possible those uncertainties.
can you tell me what was involved in developing the AI classification system?
Yeah. So I can go ahead and, take a lead on that. So when it comes to the AI classification,
we’re kind of talking more about the software side, right? So, we’re going through the route of making the app.
So EarthLens, is going to be an app, that works with the sensors and cameras to look at land in real time.
Basically, it’ll take the information like the vegetation, soil and water, and then it will use AI to give you, like, the insights right away.
So, when we were using the GLOBE Observer, we had to, use like these little scales and we had to kind of like guesstimate.
What percentage of the photo was, let’s say water tree or shadow or grass and whatnot.
So instead of waiting to process all that data later, you can see it right when happens.
Right. So, I really wanted it to be easy to use, so you don’t have to, like, be a scientist or an engineer to understand what’s going on.
Of course, that’s kind of the main point of citizen science. Yeah. So it’s meant to work in different places, too, whenever it’s,
farmland or wetlands or even urban areas so that the app can adapt, to different environments.
The coolest part is seeing the data, like, actually turn into something useful, right? So I’ve learned a lot about how the softwares and the sensors
and the AI can work together. Of course, it’s still in the development. Preliminary stages, right. Because it’s not super easy to do that.
But, also figuring out, what’s realistic for a tool like this and what it can actually do.
And what’s better to simplify it, right. So it’s been a fun challenge to make something that’s powerful, but, also very useful for citizen science.
Yeah. And, you know, developing an AI tool is, you know, it’s more doable than ever,
but it really is, kind of a space that AI hasn’t,
fully impacted, I think, in the way that it could. And so, you know, all all of the challenges that come along
with training a model, validating data to train it on, is is, you know, one of the probably one of the biggest challenges,
one that having to face, throughout the course of this project. But we think it’s very much worth it for the end goal.
So that’s one of the things that we’re excited to work on in the future. Yeah. And while we’re on the topic of both challenges and AI, I think
which is really important to note that one of our main goals was to also keep the whole thing really affordable.
And that means making sure that all the computational power that it would take is kept to as little as possible.
So we have to sort of balance how heavy of an AI model we can use without having to make sure that,
oh, there has to be like a big fat computer onboard the drone or something.
what particular say NASA missions would you be comparing the data to? If you’re or are you comparing the data to, if any?
I believe right now, in our adopt a pixel, in the adopt a pixel methodology
like chart that we’ve been using so far in this internship. I think we compare Landsat and Sentinel data
specifically to, what we’ve observed in GLOBE. And I remember in our poster, we proposed adding an additional column
with the, with the photos from down, another one for the up photo, and then like a bunch for those graphs individually.
So it could be seen as like an expansion to the currently existing, like comparisons that are being done to Landsat and Sentinel data.
So, so far I was going to say I completely agree that we we had a few conversations
with people at the AGU conference that brought up, similar topics about,
kind of thinking about how EarthLens itself could expand to slightly more than just land cover, observations.
And I think that integrates nicely, comparing a lot of different,
sources of satellite data. And also it sounds like STELLA has a few,
a few more really cool tools that we could work with.
can you tell me directly, how did you divide responsibilities among your team members, and what was your specific contribution?
So you can go, one at a time, if you don’t mind. And, Sam, sincerity like on screen. Or at least you’re in my widest view here.
I can go ahead and start with yourself, if you don’t mind. Yeah. So, before CES, before this project,
I, really can’t say I’ve had that. I had that much technical experience.
Experience? I’ve learned a lot throughout this project. I do have, some pretty significant experience
working with the Smithsonian and the communication of science.
And so, a lot of what I contributed to this on, on this project was,
kind of a how to, talk about how we are kind
of converting from GLOBE to what we think could be a better system. And then also
kind of trying to optimize for citizen science is sometimes, you know, when you have these grand ideas for projects, it’s easy to let them,
slip away from you or become much bigger or more expensive than you really have the capacity for, than what’s even intended.
And so I have found a lot of joy and, helping to kind of
stay on track towards a citizen science goal. I think in terms of division of labor,
like we were talking about earlier, I definitely had, limited technical experience.
Nandini has had, a good amount, but none of us have had an exorbitant amount of technical experience.
And so we kind of just fell where our interest lied. And, you know, we’re,
you know, it’s a small team, so everybody gets a bit of a little bit of everything anyway, but,
you know, we we all have our unique passions and I’ve worked hard to integrate those into the project overall.
Fantastic. Neeve. Yeah. So, as Sam mentioned, like, I was kind of,
going into that little bit of everything type of flavor. So I personally have experience with both
the hardware and mechanical engineering as well as software and specifically and how like both of them can be integrated.
So I helped sort of design and plan how and what the drone could do,
as well as how we would integrate that with some specific systems
that our AI and software models could use to make sure that the entire system
solves the problems that we want to, while making sure that everything’s reliable and affordable.
Fantastic. Nandini Yeah. So,
I definitely agree with Sam, and you’ve talked about how we kind of did go where our interests, lay in the very beginning.
We started our working group is actually about ten person group. And, the very first thing that I had to do
so that we could get some progress in deciding what projects we would do was, you know, introduce ourselves and introduce what we like to do in a technical standpoint.
So we had a little Google document with a list of all that and all the ideas. And that’s kind of how we split which people went to do our friends,
which people went to do a different AI based project and that kind of stuff. I would say in terms of vision and division of labor,
I always liked the hands on building component the most. I, I mean, I was I’m grateful my parents have allowed me
to take over my half my basement and, build, I guess whatever I need.
I used to do a lot more art down there, but I end up switching it into more of a little basement lab setup. So,
and, you know, I want to go into mechanical engineering. The physical hands on stuff is, what, my favorite part of it.
So that’s why I decided to build the drone itself. And, also, like, I have familiarity with being up at
like 2 or 3 in the morning and saying, guys, this thing’s finally work. So I’m sure Sam, Jordan and Neev remember the, every so often, like a 1
or 2 minute video of like, demonstrating something, working in that kind of stuff. So I mean, I feel like it, I felt what would be the highlights of my day,
I’m very happy to hear that it’s it’s so. Yeah. Fantastic.
And then Jordan. Yeah. So, so before EarthLens. Right. And I also have to say Nandini. Yes.
I did love watching those videos. But, before working on EarthLens, you know, I also didn’t have much
hands on experience with a project at this caliber. So I know I had done some programing and,
but of course, again, it wasnever at the extent that we did it with this project with EarthLens.
Right. So, specifically my role in EarthLens was really focused on the application itself.
So, I worked in designing how the app processes data from the sensors in the cameras and, you know, turns it into real time insights.
About land cover, of course, going back to things like vegetation and, soil and water, we also tested a lot of troubleshooting
to make sure the system was reliable and, accurate. So, yeah, I have to say,
I think the thing we’re most proud of was how we all grew in the process. So, from almost no experience to a project
that caught a lot of attention at that, conference. Right. And it taught us, you know, how to build
from nothing to making such, we took a big stride, I have to say.
What do you think GLOBE added to the project? I’d honestly say without GLOBE there is no project because that’s what really gave us
the foundation of what already exists and what we want to improve upon. And from that introductory sense, I think GLOBE is phenomenal.
And what it does, it’s a very accessible technology, and what it means to do is to get that scientists,
if like they get citizen scientists activated and mobilized in a way to give our actual professional scientists the data that they need.
And what we just saw was the opportunity to make that whole system better and more optimized.
And that is in no way is they saying that GLOBE is not doing what it was meant to do.
Yeah. And I also have to say, you know, science is in science without problems, right? So of course, using the GLOBE, it was a really strong introduction
to remote sensing. I personally had no idea what remote sensing was.
And then we, of course, we got into the SEES program. We got accepted into the Earth systems explorers.
And they taught us about, you know, the GLOBE app. And working working with the GLOBE app taught us, you know, to collect consistent and accurate data.
And of course, speaking from my experience, doing all that data aggregation and all that, it took close to eight hours in total.
And then I took all my photos and I was praying that they all, you know, saved to the database.
It was 216 photos I took. I lost 202 with the GLOBE Observer app.
But of course, it again, it is phenomenal at what it does. But it was just a little bit of a misstep.
Gotcha. Yeah, yeah. And it it showed, you know, it got your hands dirty.
You had to go out and do the field work and you had to do the ground truthing. And it was a, it was a
messy process of how to, you know, compare what you’re seeing versus what sensors orbiting the Earth are telling you.
And it’s great. You know, highlighting that friction.
We just think, you know, there the there’s a system that could improve upon what GLOBE has already been able
to do for citizen scientists that we can pursue.
are there any other skills that you would like to mention that you developed through the project that you haven’t already mentioned?
I’d say personally, for me, one of the main soft skills that we were able to build was definitely like working with a team.
I think this was a big emphasis that our mentors at the SEES internship had given us, and they given us resources
like their one team building module and how to do research in teams. And I feel like
it was definitely really important for us being spread out all across the country. But the fact that we were still able to be so productive and have that sort of bond
in that connection between us to make sure that we were all on the same page. You know, those videos that I was sending, like, that’s the type of stuff
that makes us all know what’s going on and be, like, happy with what’s going on and make sure that we’re all communicating perfectly.
Yeah. And I’d say also the skills to take what’s happening internally
in our team and be able to communicate that externally. So from, you know,
Nandini building the drone and Jordan building the app. And so collaborating on,
you know, how to quantify, results from the, the research we did with GLOBE and comparing what the drone gives us.
Like that’s all the technical stuff that that goes on inside. And then being able to
put that all down on a poster in an abstract or in a paper and put that out into the world for people to get an idea of what we’re working on
is it’s it’s a lot of abstraction. And that is a skill that we’ve definitely developed through this project.
how do you see EarthLens being used, say, for agricultural monitoring or environmental research?
How would you like to see it use as well? I mean, I would say the way I envision EarthLens,
especially in these earlier stages right now, is, we plan to put together, I’ve started drafting this, a guide similar to the inspired by the STELLA
Guide of components of how to build each, one of these drone systems and all that kind of stuff.
I mean, now, since each year we’ve started communicating more with researchers across the country who have expressed interest in it,
and we hope by getting it into a larger audience like that, continuing that integration with STELLA, we could have it be something
that a hobbyist or classroom, or that kind of stuff. People could actually start building these and using them in their own
communities, kind of like, kind of like how we use GLOBE. But I think the benefit of it being a drone is that as we adapt it,
to be able to fly higher and have a bigger field of view, it’ll be useful for maybe if there needs to be a weekly observation
of certain fields or agricultural areas, like you were saying with the coasts earlier and that kind of stuff.
I think it could be useful. Starting from this research perspective and continuing
from seeing like environmental stability and things like that. So, yeah, yeah, just
just to add to that, you know, you hit on the agricultural piece a bit, but the the sheer advances in technology that have,
occurred in agriculture over the last few decades are amazing. And it’s what allows us to feed, you know, the,
the modern population of the Earth and a growing population of the Earth.
That’s really important. And technology is one of the main ways we’re going to be able to continue to do that.
The way I envision EarthLens kind of being able to do that is right now you have, you know, you can survey your land to to an extent
and you can use the satellite data that’s made available to farmers to, you know, inform
the parts of land that tractors go over to fertilize or the different seeds that are planted on different portions of land.
And those are all based on satellite data as it is now. And, you know, in the research we did in the seed
program, comparing ground data to satellite derived data, we saw that there are inaccuracies in that seasonal inaccuracies,
spatial inaccuracy, inaccuracies, temporal inaccuracies. And if you have if you have something like EarthLens that’s small,
that’s modular, that’s easily deployable, that’s personalized and easy to use. But that still turns out quality data.
More immediately, you can, you know, imagine a farmer launching a few EarthLens drones and having them fly over their field
like Nandini said once or twice a week and get maybe even a more detailed image of their land than they would for modern satellite, data.
And so there’s a I really like this question about how to use EarthLens or similar technologies in agriculture.
building off of that, have you, played around with any vegetation indices since you’ve been, doing, or at least thinking about,
like, plants and agriculture and land use and all that? We’re not fully there yet.
Right now for this first iteration. Right now. I mean, right now we’re working on more mechanical challenges. Like, for example, a big thing is that the light bulb battery
for example, to power this is pretty heavy. It’s pretty heavy. And it’s time. And, right now, our maximum estimated flight time,
I’ve been working towards is probably about 20 to 30 minutes a piece right now. Now, we ought to work towards extending that or making
that battery system more interchangeable. And then in that next level we would integrate, though still, like the air quality systems and things like that.
Thankfully, our flight controller gives us things like accelerometer and, you know, other specific motion and gyroscopic related data.
But I think as we keep adding on different layers, we’re going to start, you know, maybe seeing if we can train it like, as Jordan,
you were saying earlier in the AI standpoint towards data sets with healthier or non healthier vegetation,
knowing what to flag in an image and that kind of stuff. Yeah. And I have to say, I also I really like the question
when you touch based on the, agricultural side. So I am from the Rio Grande Valley and, Texas, and it’s a region
where agriculture plays a very important role, of course, not just to the economy, but to the community as a whole.
So I’ve seen firsthand how much farmers rely on knowledge of the land, you know, the weather, the crop cycle to make, these decisions every day.
And that’s, of course, where we see EarthLens making that difference, because it can collect data from sensors
and, you know, cameras and processes, as, Sam had explained. So you know, coming from the region so connected to agriculture,
EarthLens has been exciting, of course, because it can, really improve this, agri agriculture that we have here in the Valley.
Are you inspired by any of the, different, NASA missions and Earth science that you’ve, you know, sort of, probed at?
And then which ones have you really taken a look at? Is it just Landsat or are there others out there as well?
And if you are doing Landsat, you know what? What might have inspired you with Landsat as well. So go ahead.
Yeah. So we’ve we’ve worked with quite a few different earth science projects, I think from the original data sets
we were comparing our ground truth data to, we were pulling from Landsat. We were pulling from,
a few different data sets, with Dynamic World and, yes, array and world cover and,
a few different tree canopy data sets.
And then we also, you know, again, worked with Landsat time series data, which was cool to interact with.
I for me personally, I wouldn’t say there’s one specific, Earth science project, that I’ve been,
you know, singularly inspired, but inspired by the we did hear from one of the main engineers who’s working on the NISAR mission,
which is a very newly launched telescope that uses some really cool technologies to get information you never seen before.
But I think in, in when you take them all together and you get, you know,
a more complete image of Earth systems and science on Earth, that’s kind of the part that truly, has inspired me.
You know, the goal of NASA to continue that kind of research, I personally feel is, is very valuable.
And so that’s kind of what’s inspired me throughout all of this. for it. Now the only thing right now least existing, the only thing we’re
comparing our global observer data to is that kind of Landsat and Sentinel data. And as things continue to change, I mean, having satellites that are is in the hand
of, labs on their own or hobbyists and that kind of stuff. Still being able to contribute some to some sort of shared thing,
maybe like an ArcGIS layer or something like that gives us, you know, another thing to verify against.
And I think that’s, definitely a motivating factor to keep using that
in conjunction with the satellites while we, you know, while we can and, allowing that to be a technology that continues to evolve
alongside it. what would you like to develop further if you could continue this project?
And do you have plans to. Yeah. So I’d say, on top of, like, that whole AI and automatic classification goals
that we have where we wouldn’t have to manually classify each pixel of the data that our drone takes, the AI would take care of that.
On top of that, we also have an idea of perhaps adding intelligence to the flight patterns of the drone So in a way,
making it an automated flight, which would make the entire, system a lot more efficient and make it easier for people like average
Joe’s to be able to use the entire system. And I’d say as as EarthLens as a whole, like, scales up,
there’s a lot of potential for it to integrate with external platforms that already provide data sets and have this sort of connected
infrastructure, which would just make the data that EarthLens gets. And people using EarthLens have been taking and putting it out there
for a lot more people to use, and having a lot more eyes on that same data.
Yeah, I definitely think that’s a very great way to keep progressing it. And I think from the hardware perspective, I think, like I was saying before,
integrating that carbon dioxide centered, sensor and integrating some of the things that are being added in some of the new STELLA, variations
would be something that we plan on implementing back into drone and automating flight patterns is actually, the reason why we switched to that idea
of, flight software. So that’s the easiest way to go about that in the future. So I think it’s definitely, good that we’re starting to explore
these possibilities of what we’re going to do in the future now, so we can kind of make those steps as easy as possible, like starting
from where we are today. But we definitely all have plans to keep continuing this project, at least for the next couple of years and that kind of stuff, because now
we see how valuable a continually updating or continually improving, application or system like this can be. So
yeah. And then any, any kind of tapped on it but earlier.
But the end goal is the citizen science access and the whole point of making tools
available for average people to collect scientific data is to make that
data available to researchers who, you know, we’ll know what to do with it, know how to draw conclusions.
And so we hope so. Also continue work with EarthLens in terms of,
how to get that data out to researchers and how to allow users of EarthLens to make their data accessible.
The app is the main way we want to do that, but also we hope to maybe one day have enough data to make a meaningful,
geographic data layer on some other software like ArcGIS or something like that.
how does this experience influenced your career interests or academic plans? So, like I said
earlier, the, the SEES program, you know, it was my first real kind of technical and scientific experience.
And I definitely want to continue that,
I want to study engineering in college, and I think I’m going to try and continue with that. I’m planning on going to a school
that allows you to do a lot of interdisciplinary study. And, you know, being involved in EarthLens is also,
generating a lot more curiosity for me about Earth science and, kind of the associated
technologies and space that, help us conduct our science. So I’m really excited to,
continue in that sort of direction. Yeah. For me, I’d say the process of developing EarthLens
and the SEES internship as a whole, it’s kind of me, made me firm that I want to keep being like this sort of inventor.
I’d say the fact that we were able to design and create something that didn’t exist before and this like, completely our brainchild, is so cool.
And that kind of motivates me to keep on doing that sort of tinkering, that sort of problem solving.
And especially in this, like using this interdisciplinary nature, but making sure that we’re grounded in solving real world environmental issues.
Really beautiful. Yeah, I would say for me, it’s made me more solid in my decision to pursue mechanical engineering for my undergraduate degree.
But I because, you know, I’ve I’ve had such a wonderful time, integrating this building with so many different other aspects of it.
And, like Sam, I mean, I think it’s made me a much more interdisciplinary builder, considering my prior experience is more aerospace connected.
I think that exploring Earth science and more applications here, it’s made me more willing to explore,
all different things under NASA and scientific purview. And, I mean, I plan to continue studying, hopefully
in grad school, neither aerospace or nuclear engineering. And I feel that EarthLens and this experience in Earth science
has given me, the willingness to keep exploring in that regard,
you know, and keep seeing which type of engineering is close to me at that point. Yeah. So, I totally agree with what everyone else has said.
I am a junior, so I still have, you know, a year to think about, at least a year to think about it, but not a lot of time.
Right. But working on EarthLens, you know, through NASA SEES you know, really shaped my interest in what I do want to become and that is, aerospace engineer.
So before this, I hadn’t, of course, I have already said, I hadn’t had much experience with projects at this scale.
And, you know, SEES exposed me to that. You know, the real world engineering challenges such as, you know,
collecting aggregation and analyzing, complex environmental data, you know, designing systems to process that data efficiently, too.
And, making decisions based on the real insights. Right. So I think the experience also taught me how to approach problems method.
With the method. Right. So to adapt things that didn’t go as planned and to think critically about, you know, how technology,
can be applied to solve, big real world problems. Right? So, participating in SEES influenced my goal of going to UT as well.
UT Austin. It made me realize what, I want to be in an environment where I can work,
you know, with great ambitions, you know, impactful projects, and continuing to grow as a thinker and, of course, a problem
solver. what was the most rewarding aspect of working on this project?
You know, like, what do you hope people take away, from learning about EarthLens? And is there basically anything else about the project that you’d like to know?
So, some sort of combination of those three?
I think, you know, these have been pretty comprehensive questions. You talked a lot about what EarthLens is for and how we hope it gets used.
If, you know, if you if if, you know, God forbid, the projectnever gets finished
or if younever get your hands on an EarthLens drone.
The thing I would hope that people learning about it take away is just that, these things
don’t have to be, stagnant. The way we study Earth can evolve with technology.
And I think it should, because, you know, we have better ways to collect data,
better ways to interpret that data, better pipelines for that data and more data
and all of that in in together just means we get a better picture of our Earth.
How it’s changing. There’s a lot of ways that influences policy economics.
And, you know, just in general, we hope EarthLens gets people excited
about studying the Earth for change. Yeah, I’m 100% agree with that.
And I would say, I think a big thing that I guess anyone, especially maybe high schoolers or college students or people newer to these kinds of things, can,
I guess, learn, is that, I mean, we all came with relatively small amounts of experience in here and kind of used whatever we learned through this.
Internship program to formulate this kind of mission. And it felt it felt nice to kind of, I guess, like,
as you said, to be a group of, you know, inventors in this regard. And something we weren’t as familiar with.
So I think it hopefully should encourage to keep learning and I guess keep tinkering about this kind of stuff, considering, you know,
I mean, you can you can build a drone that can do these kinds of things in your own basement. You know, there’s there is more capability to do these kinds of things
outside of the traditional laboratory or, environment.
That’s typically seen for more professionals or more for adults. Yeah, I’d say personally for me, touching on like that
most rewarding part, I’d say some of the most rewarding, like emotions that I felt were when we were all on call
and we’d have like, we’d be throwing out different ideas and then as soon as we’d hear one, then we think of one that stuck.
It’d be so rewarding just to feel like, hey, you know, I can see this. I can envision this actually coming into existence.
And I’d say one of the beauties of like, the whole process of EarthLens in general
is that in a way, it has like endless possibilities, right? Like so far, how we’ve thought about it.
We have like a few different ideas that you can go into use here. You can go to use there.
But at the end of the day, we’re still high schoolers. And that’s one of the goals we have with reaching out to experts in the area
like Michael Taylor, like people at NASA, at like for over at different universities,
because they’re going to give us so many new insights about where we could take this, what we could add and how we could keep building.
And for me personally, right, I think the most rewarding experience of working on this project was,
seeing how we could develop, what we could develop can actually make a difference. And again, reiterating we’re all just high school students, right?
I mean, we made this project from scraps, right? So one thing that really excited me,
of course, was also the power of citizen science and how a tool like this project can let anyone contribute to,
you know, collecting data and understanding environmental data. So it’s amazing, I think, that this project could really connect
communities, you know, students and researchers across the country. In terms of goals for EarthLens, I really hope it continues to expand.
You know, I think we all really think we’re just gonna, you know, keep driving, right? Of course, we’re from all literally all four corners of the United States.
Right. So, sometimes there’s a little bit of a challenge in getting, communications,
but we work it out with our, with our, time zones and stuff like that. Right. But, I also wanted to take a moment to, thank NASA SEES right.
And of course, you, Michael Taylor, for allowing us to, present this project.
Yeah, it’s it’s been an amazing opportunity. You know, being part of that program was a huge learning experience.
And, it inspired all of us to keep exploring, right? Building and contributing to projects that, you know, have meaningful real
world applications.

Related Links

|

Related Posts