Embedded - 411: Batteries Get Upset

Episode Date: May 5, 2022

Ethan Slattery joined us to talk about animals, animal trackers, and how they work. Ethan works for Wildlife Computers. They use the Argos Network for data transfer. He was previously at MBARI and worked with Engineers for Exploration as an undergraduate. Ethan is also known as CrustyAuklet on Twitter and Github. He also has an Instagram page.  Things mentioned in the show you might want to know more about: Nautilus Live is a streaming YouTube channel from an ROV exploring the oceans. They have periodic dives where you can ask scientists about what they are seeing, while they are seeing it. Watch discoveries happen in real-time. Or watch the highlight reels on YouTube. Ze Frank also has a YouTube channel about animals called True Facts that it is … not as scientifically minded. And sometimes NSFW. Start with the True Facts about the Ocotupus.  (Note he did a parody of a Nautilus Live dive).  The Genius of Birds by Jennifer Ackerman Penguin, pangolin, whale shark, weta, you might have heard about those but what about the cassowary? In-depth documentary video, people on the internet are idiots video, and Wikipedia. Transcript

Welcome to Embedded. I am Alicia White alongside Christopher White. I like animals. I really do like animals, all kinds of animals from sea slugs to the pterodactyls that are secretly called California brown pelicans to, I just like all animals, okay? And so I am super excited to be talking about animals this week with Ethan Slattery. I think there's going to be some technology too, but who cares about that? Hi, Ethan. Welcome. Hello. Could you tell us about yourself as if we met at the Monterey Bay Aquarium Research Institute summer open house. Say hello. My name is Ethan Slattery, and I'm an embedded software engineer.

I currently work at Wildlife Computers. And I guess depending on the context, I'm not sure it would come up at MBARI. I might mention past projects such as the Vaquita, or I actually taught Photoshop and stuff like that for a few years and was in the military for a while. And what does Wildlife Computers do? So Wildlife Computers, we make wildlife telemetry tags,

and that's all the devices you might see on Discovery Channel or other places that track sharks, whales, other marine life as they go about their business in the ocean. Okay, we want to do a lightning round, and I know that you listen to the show, so you're familiar with this.

Are you ready? I am ready. Have you ever pet a penguin? I have not pet a penguin myself. My wife has been to Antarctica and worked with penguins. What about a pangolin? I have not touched a penguin myself. My wife has been to Antarctica and worked with penguins. What about a pangolin? I have not touched a pangolin. Platypus?

No platypuses. Cassowary? That would be a little frightening. I don't think I would like that. We just discovered cassowaries recently. They are amazing dinosaurs. They are amazing dinosaurs. They are amazing dinosaurs. It's dangerous.

All right, all right. You got to have an octopus. I guess I have interacted with an octopus, yes. Whale shark? No whale sharks. You want me to continue this long list? Okay, what about a leatherback turtle? Leatherback turtle, no.

Okay, what is the most interesting animal you've ever pet? This is the better approach. That's hard to really nail down, I think. Can I say two? And that's the crested auklet, which I am sure will come up again. But they're these little, somewhat penguin-like birds, but they smell very strongly of tangerine. What? And they have fluorescent bills.

And then elephant seals, of course, are always at the very, very top of the list. And I worked with them for a year or two when I was in school volunteering. And I mean, they're just enormous animals with just, I mean, the list can go on and on about what makes them interesting, but they're just, you know, present day megafauna that just are just there on the beach so we have uh an elephant seal sanctuary sanctuary uh here in santa cruz which i believe is where you interacted with them and one of the most amazing things in the tour was the idea that they are slightly bigger than an old style vw bug

yes i think the very first time i interacted with one up close uh so i was volunteering with the lab there at uc santa cruz and one of the main jobs was to go out every morning very very very early and they put these flipper tags on them that are kind of the same plastic tags you'd see on a cow ear but you can depending on what angle it's at and which way the spike is going through where it's top to bottom or bottom to top relative to the fin and the number on it of course and the color they can identify that animal and exactly when it was tagged and you know you recite it year to year well the researcher was like go figure out where the tag is on that animal over there and i looked and i what animal is it by that huge folder and that that was the animal

and i had to get but did not get too close but a little closer than i felt comfortable with that first time uh back to lightning round do you have a favorite fictional animal or cryptid uh should have thought about this a little more but probably you know giant squid yeah okay they do exist a little bit they sort of exist yeah sort of exist it depends on how giant the giant squid is right where it goes from fictional to non-fictional. Okay, I guess there's a follow-up to that then. Is Bigfoot real and are you tracking it? I can't talk about that.

All right, let's go back to crusted auklets. Sure. Are they auklets that are crusted? I mean, it's not crested, which means something else. That means it has, you know, flair. But are they crusty? No, they are crested. They have this mohawk.

Not really a mohawk. It's like just a tuft of feathers on the front of their forehead that kind of goes up. And the more it goes up, the sexier that individual is. And that's male and females, I think. Yep. Oh. They're not like Krusty the Clown. So Krusty is, I think, you're getting confused with my name often, is Krusty Auklet. And that goes back to a very specific story and a very specific Auklet I interacted with.

Was it Krusty? It was. So when you walk from our campsite to when I was in Alaska for about a month on this island in the middle of nowhere, tagging these animals, these crested auklets, you have to walk from your campsite to where they nest. And when you're walking to that area, you go over this field and they're not very good flyers. They fly like bricks. And so they're trying to get from other nest sites that are higher up the mountain that are too far for us to actually go to, down to the ocean. And so they just kind of do the bare minimum over the ground. And if you look up at them, it's funny because at any moment in time, it just looks like someone just threw a handful of them in the air because they're not flying straight and they're not flying well. These birds look ridiculous because they have this little floof.

And when you say crest, and like, okay, what is a crest? A crest is like a bird mustache going vertically instead of horizontally. And they look, I mean, there's the old cartoons with the quails that had to blow their their crest out of the face they look like that except their eyes are tiny and beady and really kind of dumb looking so they're very optimized for diving down deep into the ocean so those eyes right there they're diving several hundred meters down in the ocean to get krill and so their eyes are when they're up with the surface are not as dilated as they usually are right there they're not used to that bright light. And that's also why they don't swim, fly very well, right?

Because they're optimized for swimming. They do swim with their wings and their feet. Ah, so they're kind of halfway to penguin. Why do their bills fluoresce? And how did you find out? Did you take them to a rave or something? So it's very hard and speculative to say why, right? I don't think that's a question we can really answer scientifically.

But the how is that I had taken my dive light. So I'm a diver and I had a UV flashlight. And I had taken it because it's a pretty good flashlight. And the UV, it's optional. But we went outside at night and I shv it's optional but we went outside at night i've shown it a round and the whole beach just kind of lit up with this bright orange beaks glowing in the dark and from there we just kind of wrote a paper about it and we had a bunch of skins to study back in storage where i could verify it and look at different species and And yeah, it was really interesting. What's the purpose of that, do you think? Or do you know?

Again, we can't really know. I mean, I can make a lot of assumptions. Maybe it helps them look more attractive to mates, or maybe it's just accidental because they eat krill. And it just happens to be the element that the krill give the color, right, to their beaks. Because those beaks do, the bill plates fall off out of breeding season. So they will look not quite as cool when they're not breeding. Does this mean that they probably can see UV?

So I think just to say that they'd have to dig into the eyes, but I'm not sure if anyone has done that. But there are many birds who do CUV, right? But I don't think you can say all birds. And again, I should probably interject that I am an embedded software engineer, so I have many friends that are biologists, and I spend a lot of my time around it, but nothing I say is official. Okay, okay. Back to the engineering then.

I'm not a birdologist. We have some listener questions, and I'm going to start with those instead of end with them, because a lot of people had very similar questions about trackers and animals. And what happens when an animal that has a tracker gets eaten? You're just starting with that one, huh? I think it's what everybody, I mean, we had a whole bunch, everybody wanted to know that. Our audience wants to know. So I guess there's, so when a tracker is eaten, so you're saying when an animal with a tracker on it is eaten, there's two categories in my mind, and that's if it's intentional or not.

And so I'll start with the unintentional. And so I guess there is an interesting story there. Unintentional predation? Yeah. So we have seen that in data. And what's interesting is that we very rarely, and we can get back to this in some of the challenges of the engineering, but we very rarely actually get an archive back. So these are data loggers for all intents and purposes, but we very rarely get the actual

data archive back. We process and transmit that data over the radio waves. But in that data, we can see that there are data sets where you can see that all of a sudden the behavior changes. So maybe the animal's use that it's on is diving and interacting at the top layer of the water or the first couple hundred meters. And all of a sudden, it just starts going really deep. That's probably indicative of it being a different animal.

And sometimes associated with that, the temperature will normalize. It's in the stomach. And the light goes away. And all those things kind of indicate that it's been eaten, which is pretty cool to see in the data. And then do you ever see data when the light comes back? Yeah.

And I guess the question is, which way does it come out? A new way or the way it came in? And I don't think we can actually know that for the data, that we don't get stuff back, right? But I would guess back out the way it came. And that's because the intentional ingestions, that's what usually happens. And so that some researchers have used RTAGs to intentionally intentionally feed them to animals and then they use that to study i think what they're usually hoping to see is right uh how often and what they eat and infer that from temperature changes and stuff over time why do they get barfed up i'm sure a big uh electronic or we hope they're not

too big but right they're not the best food. When was the last time you swallowed a Fitbit? Follow-up dumb question that I have to ask. How many levels of recursion of tracker eating? I love the question, but I have no idea. That's a good question. We should look into that if so many questions about like okay so people put the trackers on the prey food and then feed them to the predator food to find out what and what kind of prey and predator are we talking about

i wouldn't say that they put it on a prey item and then feed that that probably would not be ethical um i don't know for sure but i suspect they either feed it directly or you know put it just in the treat of some kind or some kind of food item that's not alive um but yeah usually it i think there's many reasons but the reason that is in my head is they do study you know how often something is going to eat because when you know how often things are eating you're studying energetics and stuff like that so that's important to know how well something is eating

if it's doing well or poorly in the environment from an engineering perspective it's cool because I did work on a device once I have this tag on my desk for a couple months that had been inside of a shark for a while and it definitely looked like it had been inside of a shark it looked like it had been inside of a shark for a while. And it definitely looked like it had been inside of a shark.

It looked like it had been inside of a shark? Yeah, it was a little acid etched and had some gnaw marks on it, small scuffs all over it. Just can imagine that, you know, I like to think in my head where it had been. And that was cool to me. How much do you have to worry about the durability of the trackers? And does that depend on the environment they're going into, water versus land-based? So how much do you have to worry about the durability of the trackers? And does that depend on the environment they're going into, water versus land-based?

I guess, so to clarify, so where I currently work, we pretty much only with, we have some exceptions, but it's mostly marine. So we don't do any land animals. I'm sure there's an exception I'm forgetting, but primarily marine stuff. And yes, durability is very, very important. So most of our tags are rated to go down to 2,000 meters of depth. And so these are in basically solid epoxy blocks of different shapes. With no air bubbles. No air bubbles no air bubbles yeah so i mean this is outside of my

expertise but the mechanical engineers that i work with uh basically do magic as far as i'm concerned they're amazing and so these different shapes with no air bubbles epoxy and uh marine foam it's called of different densities right that can go down to these different depths and calculate all that out so that certain tags pop off which we can get into and it has to float to the surface right and just survive all that extreme temperature and pressure changes when i was working a little bit with ambari in order to get things to float they had to have like an explosive release of air how do you how do you get recoverable tags or even tags to go up so that they can talk to your satellite system yeah so it's just this marine foam and again that's the mechanical engineers that a little bit outside my realm but we don't have any of those explosive

um i guess floats and that probably does limit it a bit i wouldn't want to feed one to an animal but so the the tracker goes in or on and then um it's just that the float is light enough that it eventually surfaces yeah so they are attached um so it's at the float there's a kind of a fine balance again probably mechanicals do all this calculations but to where it's attached to the animal and it does not impede on their movement or behavior too much, right? But when it releases from them, it can float to the surface. And so it is permanently, or semi-permanently

attached to them. And then there's a metal pin that burns, not burns, but dissolves in the water when it gets electricity applied to it. Is it electricity or is it just rust away? No, it is electricity that causes the corrosion of that pin and causes a release. And then that attachment device will stay on the animal until it kind of gets moved out of the animal

through their natural healing processes. Can you say, okay, release now, or is it usually a timed thing? Oh yeah, it's timed, and so we can get into that, but we have no communication with these devices when they're in the field. Until they're released? Even then, it is one way. Right, you can't talk to it. It can't talk to the devices. And how does it talk to you? So it's floating on the top of the ocean, I guess, at this point.

Mm-hmm. Having been released from whatever animal it was on. Yes. And then what happens? And then it, I like to say it screams into the ether, hoping we'll hear it. Because we use a satellite system called Argos. So anyone who does ocean stuff will definitely recognize that name. It's used for like buoys, NOAA, stuff like that.

You know, all those oceanographic buoys use it. But although there are different kinds, there's different protocols they have. The one we use is one-way. And so it, yeah, it yells up to the satellite. These are these polar orbiting satellites. And so we just up to the satellite, these are these polar orbiting satellites. And so we just have to do a lot of work to make sure we transmit the same data enough times that probabilistically it will hit a satellite as it goes over. But satellite data is very expensive,

so you don't want to transmit too many times, but you do want to transmit enough it gets heard. How do you balance that? So it's configurable because the customers are paying for that data. We are not paying for that data. But yeah, it's configurable how many times it will attempt to transmit a single message. There are ways if you kind of know where it is in the world, you can set those configurations up because you know when the satellites will be there, right? And so you can say, don't transmit during these hours and transmit during these hours.

But if you're kind of the most generic tag where you don't know where you're going to end up and what time it is or anything, then you kind of just have to transmit a lot and a lot of duplication and hope it all gets through. There are some animals that stay pretty much where they are. They're maybe not territorial. Do ocean animals, are they territorial? I guess octopus are. Maybe that's a little bit out of my realm,

but on that note, there are migrations that are predictable for sure, right? So elephant seals, going back to them, very predictably come back to the beach there in Santa Cruz for breeding and for molt at very predictable times of the year. Where else do they go? That's a cool story, because it's actually tagging that lets us know that. I think before, so before tagging, no one really knew. They kind of assumed they just hung out near the shore. And so, for anyone who doesn't know, they come to the shore for molting and breeding, and other than that, they're just gone. They assumed they come to the shore for molting and breeding and other than that they're just

gone they assume they were near the shore but they actually the males and females go to completely different areas to go feed and it's like hundreds and hundreds of miles out to the ocean up towards Alaska and that area I think for the females and then the males I forget exactly where but yeah really really far away

they go super super deep they're some of the deepest divers in the world yeah just tons of interesting thing that we just had no idea before tagging Really, really far away. They go super, super deep. They're some of the deepest divers in the world. Yeah, just tons of interesting things that we just had no idea before tagging. Another thing, for example, most animals will transit. So if they're going somewhere near the surface, right? Diving deep is kind of stressful, you'd assume. And they'll transit near the surface, but elephant seals will transit at depth. They like being deep.

Very strange animals in a cool way. What are some of the strangest things that have happened to your tracking systems? So one interesting thing is we had a tag on a tuna. So tuna are something that get tagged a lot. And that tuna, when we saw the archive archive and it kind of suddenly dropped in temperature. And so we got that tag back from a fish processing plant. And so what had happened was that fish was caught and thrown into the flash freezer. And then they didn't see the tag until it had went through the bandsaw and cut the tag in half.

And then we have our number on all the tags so that people find them. And it's always a very exciting day when they find a tag that we don't expect to get them back. But they find it on the beach or something. And then we can download the archive. But the fish processor found this cut in half tag. And I'm not sure what happened with the battery because it went right through the battery. And we got it back, and we actually got the air drive off that one, even though it was cut in half. Very spicy sushi.

The Argos network, you said there are different kinds of configurations with it. What kind of configurations have you used? Only the one-way or other ones? Yeah, I've only used the one-way. So Argos 2 and 3, it's called. There are other ones that are lower power, that have smaller data packet sizes, and then there are some two-way options. I think it's a future possibility,

but another consideration is that a lot of these animals, if they only surface for a second at a time, maybe one or two seconds you have of seeing the air. So you don't have a lot of time to do any kind of two-way communication. You're kind of, like I said, just screaming

out into space and hoping it's there. A transmission takes about a second. So what do your trackers actually capture in terms of data? Is it like capturing, it can't be capturing GPS location all the time because an animal at depth isn't going to be able to have that. But what do you track? So localization, or what biologists will call horizontal movement,

is one of the most important things, actually. Oh, okay. But other than that, so data is like depth, temperature, light level. And then any other sensors, like there's dissolved oxygen. There is, you know, salinity of the water or conductivity of the water. There's all kinds of cool special sensors that a biologist might give us to integrate do they ever do any acoustic kind of stuff to see oh this this animal's in a loud area with human interaction i can't go into too much detail about that but there are you know special tags for special you know special request

tags that do acoustic stuff and yeah impact statements stuff like that with environmental impact oh cool how do you do localization underwater i mean you can't use gps yeah so that's what i was gonna get to next and because like i said horizontal movement and location is one of the key things um so there's a couple different ways and one of the older and still common ways is light so just capturing the, and then we have algorithms we run on that light archive. And so if you imagine that map of the Earth with the light, you can kind of figure out

where you are latitude and longitude from the time, the length of the day, and what time actual sunrise and sunset and different parts that they were. And you can get a pretty good idea of where you are. So it's not GPS accuracy, of course, but when you're tracking an animal that's moving across the planet,

it's pretty good. And that's all you need is light data for that. There are publications about how exactly that's done, if anyone's curious. It's really interesting. Other than that, there is acoustic, which we don't really do right now, but that's fixed locations that emit

sounds in the water. and then you can try location and then gps we do actually do gps um although we have our special version of gps called fast look and so the kind of high level that i can get into that is just that we we capture the gps data the rs spectrum and then uh do special summarization and statistics on that and then send it back and the actual location is calculated back later oh interesting right the animal doesn't need to know where it's at just pick up the important parts of the ephemeris and just yeah send that back that's fascinating I mean how how much processing do you do on the tracker versus how much do you send back to be analyzed? Yeah, it's all processed for the most part.

So to go back to Argos, what would be the most common Argos channel, you get 32 bytes of data. And you actually get 31 because they stole one byte back for their transmitter IDs that they didn't make them big enough at first. So you get 31 bytes, and they have to be atomic. You can't rely on other messages making it through, right? We talked about how you have no confirmation

and you don't have two-way communication. So you're not splitting up 1K of data into a bunch of small packets and reassembling them easily. You can't really rely on that. If you split it up, you need to make sure that it's okay to have only one part of it, right? And so with that in mind, we process everything.

So it's lots of statistics, lots of biologists coming up with the algorithms, and I implement it. It's part of the job I love. You get to kind of do these high-level scientific algorithms in an unembedded context and on implement it. That's part of the job I love. You get to kind of do these high-level scientific algorithms and unembedded context on this data. I know that it's going to make it through. So you do most of the analysis on the unit because the cost of bandwidth is so high.

How do you do testing for this? Yeah, there's a lot of end-to-end testing. It's kind of one of the primary ways, right, to get it through an employment, is what we call it. But to exercise the collection and then the transmission of that data, then also breaking that up into the different sections of its lifetime. But it is difficult, right? Because also some of these devices, so we didn't

get into it yet, but the two different type, big two different families of devices is ones that collect for a long period of time and then float to the surface and then transmit, right? And then there's the ones that are on maybe something that breathes air, so it gets to see the sky on a regular basis, and so it's going to

kind of in real time create data and send it back and those ones that float to the surface are very difficult right because it might be a year or a two or two collecting data and doing nothing else and then transmitting and so to test that we can imagine is very difficult to actually get two years of data right so yeah we also have a pretty extensive simulation platform written that we use for that kind of stuff. And yeah, it's just lots and lots of testing. Both kinds of those sensors have similar issues with making them small and durable and power efficient,

which is part of making them small. What are the differences between long-term one splat of data versus long-term but lots of little data? Deciding, just design decisions. So when it's for a long, long time of collection and then kind of doing all the data at once, it's a little, you have the advantage that you can kind of look at the data collected holistically, right? You can compare days or you can look at a long period of time and do some statistics on that.

Whereas if it's in real time, you have to decide when you're going to summarize that data. And maybe, hopefully we figured out that that's the best time to do it and we don't miss something. It seems like there's a big power consumption difference, right? Because if you're coming up regularly and transmitting, then you're drawing off a certain amount of power regularly. It's fairly consistent. But with the one or two year ones,

which might be underwater for a huge portion of that, you've got to conserve enough batteries that you can actually transmit to a satellite at the end of the life of the thing. Because it's all wasted. If you don't get the data back, it's all pointless. It's an interesting problem. How do you kind of back to the testing, how do you be sure that, okay, it's up and we have enough power to transmit? So you can characterize that power for the different modes of operation and then

estimate how much time you're spending in it. But back to that battery, it's actually something interesting I learned here, and that's maybe this won't even be that much of a surprise if you really think about it, but batteries also don't like working at very, very low draw, and then all of a sudden if you really think about it, but batteries also don't like working at very, very low

draw, and then all of a sudden having to do hard work. So we're running at the micro-ramp level for a year or two, and then all of a sudden having to transmit into space, batteries get upset. So the batteries, there's not only that you have to conserve enough power to get all the

transmissions you want at the end, but you have to make sure you use enough power to keep that battery in shape. And I presume they're very cold, which they also don't like. Yeah. Batteries do not like working for us, but we make sure that they do their job. And so that goes into the testing and development of the tags. How much of the testing that happens before a product is released is unit testing versus simulated testing versus hardware-in-the-loop testing? So historically, a large portion of it is, well, simulation is a huge portion because that allows us to use historical data sets to test the algorithms and the summarization and the transmissions in the simulator.

So test the transmissions, what messages are going to get transmitted in what order and all that. Hardware in the loop is definitely something we've been doing more and more of, especially for the pop-offs. It's nice, we call them pop-offs, the ones that collect a bunch of data and then transmit it,

is because you can inject all that historical data also and then the hardware doesn't know the difference that it's historical data or if it collected it, and then you can run that in the hardware in the loop for the transmissions and the creation of the data.

And unit tests is definitely also something. I'm actually really big on testing, and so I'm pushing a lot more unit testing and stuff like that in our code bases. What is the biggest... What is the difference between the biggest size tracker, largest size tracker, and the smallest size tracker? And does that correlate with the largest size animal

and the smallest size animal? Well, yes. So there is a limit on what kind of size tag you can put an animal there's you know it's about three percent body mass depends on the regulations but it's you know that is something they worry about biologists so that's they pay attention too closely and the bigger it can be i mean the first thing i think of is the more batteries you can have right yeah so for something like a whale or something very large and heavy right uh you can you're not quite so power constrained as you are with a small fish how small can you go um so i'm gonna get wrong if i say off the top of my head but yeah

going smaller if that's an interesting answer is is always a goal. Not just for its own sake, but if you imagine scientists want to study these animals and they want to study young animals as well. But if your devices are large, then you're limiting the population of animals that are able to be tagged. So outside the scope of even my work here currently is before I was tagging or helping tag nose leopard lizards. So there's these lizards out in California. And they're very large lizards, but their tags have to be in the one to two gram range. One to two grams? Yes. And a nickel is 5 grams? So this device, and I didn't make the device,

I was just helping with the physical labor, which is fun because you get to interact with the animals, but it was pretty much just an antenna and an oscillator on like a hearing aid battery. And there's no, it can kind of change the oscillation and they can detect that to maybe see the temperature changes over the day and that's about all the data they get. Other than that, and this is interesting in the way the tags are used, is that it's just location and you have to find it. So we would go out every day

with big Yagi antennas and you're kind of doing the transmitter hunt until you find this very well camouflaged lizard in the desert. And then you sit there and watch it for 30 minutes and write down what it does. And you do that every day. You've been on a fair number of such adventures. What are some highlights? So that was one highlight, those leopard lizards in the Carrizo Plains in California. And they're really beautiful lizards. And just seeing that technique of just, you know, they're using this technology, but it's also very low tech at the same time, right, to track them down and then just watch them and write it down in notebooks what they do.

And so then in Alaska, we went to the island of Garaloi, and that was very, very cool. I thought I was still in school at the time. And it's this island out in the Aleutians, and it's about a three-day boat ride from the closest inhabited island. And there's usually no one there at all, but this research student I was working for had the permits to go out there and study these animals. And so I was hired to go, and we actually were retrieving the tags, and they were these tiny, like, underground tags you strapped to the legs. And again, they're only collecting light, so they actually were archiving, but just light data at a very low resolution, like minutes between samples, just to see where they were going because they're very small animals, right?

They're two to 300 grams. And, but they are, you know, we'd go out every day and get these tags off their legs because they, again, it depends on how the animal works, but they come back to the same burrow every year in this lava field.

And so we knew where the holes were, and you would go back, and you would reach your hands in there and grab the bird and pull them out, and they weren't very happy about it, but you'd look, and if they had the thing on their leg, then you'd snip it off, and if not, then you'd know, and again, they're very behavior-oriented. 24 hours later, their mate would be there, and that was the one that had the tag hopefully so you put the tags onto the birds and then they wander around wherever they go and you get light data and the light data tells you how far south they are and even

based on if it knows what time it can figure out longitude. How does it keep a good time sense if it's so small? Isn't there drift? Oh, yeah. So clocks, right, with the temperature changes are going to drift and the extreme temperatures these animals go through. It's going to drift quite a bit. The light-based, I don't know off the top of my head how sensitive it is, but it's not very sensitive.

You don't need second or sub-second accuracy or lack of drift. A couple seconds or minutes, you're going to move maybe fractions of a degree as far as location. So it's okay. For stuff like GPS, it matters quite a bit more to know your actual time. But at least for the tag they make for work,

we have the benefit that those transmissions. And so this doesn't work for those ones that collect for a year and pop off quite as well. But the ones that transmit on a regular basis, you have a timestamp of that transmission from the Argo system, right? When it hits the satellite. And so you can adjust for that drift. And so the birds are in their burrow.

That's not going to have any light. How much does the animal's behavior mess up your data? How dare they? How dare they? Don't they know they're being scienced? So again, it's very rough. It's over the course of a day and you smooth that. So it can mess it up and then you miss that day maybe and then

extrapolate between but i'm assuming so that those tags were from a different company and these light algorithms tend to be very proprietary even though there are you know they're published academic research on how it works but every company's uh light algorithms are very hush hush because you know we do lots of work on that and special tricks and filtering that make it work better. Is there public access to any of this data? Yeah, that's a really great question. There is quite a lot of this data available publicly. So like any scientific data, I think a lot of times they'll hang on to it

until they get their publication out. But there are a few government and nonprofit run websites where all this stuff ends up. And if you're interested in data science or visualization, that's a great resource to go to and just pull all that data and do whatever you want with it. Machine learning of crusty occlet location. Crusted. Crusted occlet. A comparison of shark and

crusted occlet locations over the last 15 years. Come on, these are the sorts of papers I want to read. What websites can you, or do you want to send me a link and I'll put it in the show notes? I can send you links for the show notes. Yeah, they're really good.

I don't know if they're in my head. There was a recent one I saw where they had, you know, you can see the whales going around the ocean and where they go. Yeah, it's very cool data. In some of the documentaries we've seen, there are animatronic robots that pretend to be the animals and they travel along like the dolphin cam is that is that real science or just entertaining tv oh uh i hesitate to make any quality judgment i mean it depends on what they get out of it right but i just know when i see it i definitely get a kick a laugh because they they don't look very realistic to me, but the animals maybe accept them as they are. Well, I mean, isn't that a lesson we should all learn? Of course.

Have you seen them in person? Have you interacted with the animatronic ones at all? I have not, no. So the wildlife tracking industry, as you mentioned, there's a lot of science and a lot of government data. So this is driven by grants and research projects. That's a tough industry to work with because their funding is sometimes odd. Does that impact your product development or are you kind of at the end of that? Yeah, it does to a small extent, at least where I'm at. So I'm kind of at the end of that? Yeah, it does to a small extent, at least where I'm at. So I'm kind of at the end of that as an engineer, a development engineer.

But I definitely see that sometimes there will be a proposal in the works and might spend some time spinning up on a new technology only to have that not approved and then kind of move on and have to switch gears very quickly between technologies and projects. And that's the biggest effect i see um and then keep in the back of your head because maybe it'll get approved next year the wildlife tracking industry is is kind of one of those industries that doesn't change very fast um but you're talking about one gram sensors and and being able to do things in a

tiny amount of space do you find that they're accepting of new technology or not um i think it's they both are and and aren't so they the r is that it's very accepting of new technology to meet any need of a special research project or special grant. And it's very quick to try out new things. And the R is, or the not is, there's also long, long-running research studies, right? And they don't want to change the technology midstream. For example, that elephant seal study, I mean, they've been going for decades. And there's like many, many other research projects that have been going for decades on different animal populations. And sometimes they're hesitant to, for good reason, right, change the technology they're using to sample their data.

Does that make it hard to continue producing stuff? Because at a certain point, you're not going to be able to obtain the same parts to make the same devices, right? Especially recently. You can't obtain any parts right now. It's leaving the current, yes, aside. But, you know, I want this 180-51 from 1995 because that's what we've been using. Yeah, I haven't seen that. I think you do lifetime buys, and again, that's a little bit out of my, I don't do the buying, but it wasn't an issue. It has come up a few times recently, of course. I's funny when they define raw data differently than I do, which is

what the sensor gives me, not the output of the quaternion. Do you have that problem? They really want raw data, but they can't afford the bandwidth charges? Yeah, I think everyone would love raw data. It's not something we struggle with, because I think it's pretty well understood that it's not just

the bandwidth charges, right? It's physically impossible to stream back live sensor readings over 32-byte packets that you occasionally get. So there's a lot of understanding there, and they work closely with us to design the packets we do send.

But yeah, I mean, I said before, everyone loves when a tag is recovered, right? Especially if it's one you weren't expecting to recover. Some are designed to be recovered if it's an animal that facilitates that, right? That comes back to the same location. Elephant seals, for example. They're always coming back. You're going to get that tag back

if it's still attached in other animals like that. But something like a fish, if that tag washes up on the shore and someone finds it, everyone gets very, very, very excited. It's true. found a a noah and a one of the uh radio sound things yeah and and it actually we found out later it said somewhere that you're not if it's been in the ocean they don't care anymore but we were so excited to

find it and we wanted to send it back and it was was very exciting. So, yes, I believe that if I saw one of your tags on the ocean, I would be like, okay, here it is. First thing I'm going to do is see if I can get the data off. If I can't get the data off, I'm going to send it to them and demand they get the data off and show it to me. First thing I'm going to do is attach it to a completely different animal. Yes. I'm going to carry it around with me for days. You mentioned the algorithms is the most fun part for you, taking what the scientists come up for analysis and implementing that. And obviously that's probably some of the secret stuff about the work you do,

but without revealing any of that, what are some of the things you've learned as you've done this to kind of fit what might be difficult computational things into these small devices? Interesting things I've learned, and I don't think it is super mind-blowing, but like the difference between a live algorithm and a batch algorithm and making the design decisions between those was interesting to me. So whether you're constantly getting the result right, or if you need a window to process that result, just the different ways to get the math

that you need to do with a small 8-bit processor or a small embedded processor to do division or trigonometry or matrix math. And do you have a lot of time to do these computations? It's not real time, so okay, well, we're going to take five minutes to do something that on a smartphone might be expected in five seconds, but we've got all the time in the world. Yeah, for a lot of the stuff, it's not real time. The biggest time constraint is the power, right?

How long you're on to do that processing. It's not really meeting a deadline to get that data out for the most part. At least I haven't hit that yet, right? Because we're talking minutes before you would actually miss. If we had a deadline, you'd miss it. Okay, so going back to your student days and the elephant seals, what's with the hats? Did you actually put hats onto seals? And if so, can I have like a hundred pictures of seals wearing hats?

And were they jellyfish hats? I have one picture. But yeah, they were not jellyfish hats. They were epoxy block hats. But so in school as my... I now have this idea of an elephant seal, a huge elephant seal wearing like one of those triangle cheese head hats. But maybe you should tell me what it actually looks like. Yeah, it's so that the seal hat, we had to have a clever name, of course.

And I think my team came up with that. And it was the SEAL Heart and Activity Tracker. Luckily, the sponsor of our project wanted to look at the heart activity and their regular activity. And so we came up with a clever name. And so we designed this kind of just data logger that tracked activity. So we had accelerometer, light, kind of the data logger that tracked activity. So we had accelerometer, light, kind of the basic sensors on their temperature.

An accelerometer is one of the big ones for activity, of course. And then a Maxim chip that was an analog front end for heart activity. It's like an ECG. And then figuring out how to attach those electrodes in the know, in the right place on the seal was fun. And gluing that to their back. Okay, so did you just like stick them on like they do with humans? I mean, these seals are really big. How do you do ECG on this?

Yeah, so the sponsor is this professor and she was doing, had done this before, right? And so I looked at different animals and their heart activity. And so she kind of knew where to put them and she was helping us a lot but you have to shave you know a little patch of fur away they have very very dense fur and then it was just kind of off the shelf heart heart rate ecg pads that we put on them and they got really good signal actually with this this chip that was made for humans as well and with the right you know processing which wasn't much it got a good reading and you do that by so there's this whole process we went off the beach and it's kind of was tied to that volunteer work i was doing uh reciting them with those flipper tags but they also at the right time of year will tag them right with the

with the commercial tags not the student tags and they also take blood samples and do some physiology work. And so we kind of piggybacked on one of those on one animal to test our device on it. And it was in a red SparkFun box with the electrodes coming out. And there's just this huge animal that was asleep because they tranquilized the animal while you were with them. Yeah, I was going to ask. You're not walking up to an elephant seal and doing anything, right?

No, because you're pretty scary uh it's kind of the senior grad students that do that and it's very impressive because they have this uh a syringe at the end of a long tube and they have to really sneak up on the animal and then get it in the right location and then back away really quickly and you know push the plunger that's the end of this long this long hose and then you can back off and dodge them a little bit until they fall asleep. I'm surprised the senior grad students do that. It's like the thing you make the first years.

Maybe they tried that and it didn't work out. The juniors do very well. Do the other SEALs get mad when this happens? It's kind of funny. They don't care too much about each other, I guess, in that way.

Without anthropomorphizing them too much. But yeah, they don't. They'll just kind of clear out a little bit. But they're very, very quick. A lot quicker than you would expect for how big they are. So you have to always keep your eyes open for a male who

don't even really acknowledge your presence. But if you're between them and a male that they don't like, they will quickly run you over if you're not paying attention. I mean, you're seriously between two VW Beetles who might at any time go in your direction. And it's not like they have headlights that you can tell which way they're going. They're really fast. Yeah. I feel like it's worse because of their density. I don't know why. It's more dense than a VW Beetle.

Yeah, that's why the senior researchers are the ones who are the experts there. They have a lot of responsibility to watch over the silly volunteers who don't have a lot of experience probably. I'm sure I was one of those when I first started volunteering. Maybe even towards the end. who don't have the level of experience probably. I'm sure I was one of those when I first started volunteering. Maybe even towards the end, I don't know.

So you spent a lot of time in the field doing stuff as a student. And now you're doing software development. How did that path work? Yeah, so when I graduated, well, before I graduated, I internships at this company I work now I work now wildlife computers and spent a summer here working for them and they decided to hire me after that and I accepted and so I moved up here and started working so it's not a very exciting story I guess but it's the standard right like I graduated and came up here and started working on the engineering side and what. And what was your student?

What did you study in school? Yeah, so I was a computer engineering student at UC Santa Cruz. And I specifically chose Santa Cruz because I knew they have this amazing elephant seal lab. The Costa lab there is kind of world known. And it was a great choice I think and they have a good computer engineering program as well and so it was kind of the perfect mix for me so your goal was to do marine stuff to start with but do computer stuff

okay cool so which of these is your favorite? pinnipeds, cetaceans, or cephalopods? Pinnipeds, for sure. Which ones? Elephant's tail. Oh, wow. Elephant's tail is one of my favorite animals.

I walked into that one. But you also worked with, worked for, vaquitas, which are cetaceans. Yes, that's true. So that's another project I did, I guess, when I was a student, and that was very early on. So I had just started school again after my previous careers. And there's this summer program called Engineers for Exploration down at UC San Diego. And it's a very, very cool program.

So it's an REU. I don't know if anyone, you've ever mentioned that before, but it's these research experience for undergrads through NSF all over the country. And so they have a specific one at ucsd that works with wildlife but also archaeology they're doing a lot of 3d scanning of like mayan ruins and the project i got onto was this vaquita project and the goal was to photograph a vaquita and so for anyone who doesn't know they're this endangered porpoise in the gulf of mexico, not the Gulf of Mexico, in Baja.

And they're very, very endangered. Like there's less than a dozen, I think, now at this point. It's very sad. And they want to get a picture of it. It was through the grad student I was working for, the NatGeo, it's kind of like sponsored by NatGeo, the E4E, and she was working on this project. And it was this, I guess you see a lot more now,

but the 360 degree kind of sports cam idea where there's a bunch of GoPros. That's what we had at the time. I kind of point in all directions and then in a waterproof housing and I was working on the software that was running on Intel Edison, if anyone remembers those. Oh, geez, yeah.

Yeah, so that was fun. It was like an x86 embedded computer and struggling a lot with trying to get Linux to be more real-timey to get those audio samples. Because we were using audio to detect their presence. So they click at a very specific frequency, which is a very high frequency, unfortunately, like 135 kilohertz.

I'm trying to use audio systems to hear that and then turn on the cameras and get them. They're the smallest of the dolphins, right? Yes, porpoise. Porpoise. What's the difference between a porpoise and a dolphin? I should have written this down. I'm going to get it wrong.

But it has to do with the shape of their teeth. Okay. So one has cone-shaped teeth and one has spade-shaped teeth, and it has to do with what they eat. Yeah. Did you see any? We did not see any. That would have been very, very special with how, not only how small they are and how rare they are, and also the water there is very murky. Like, you can't see far.

It's maybe 20-foot visibility most of the time. Did you hear any we did not oh well this was the world's most depressing segue very sad the technology on that was very very interesting but at least while i was working on it we did not get any um pictures or or hear the sounds part of the reason they're so endangered is because there's well i mean not because they're so small because they go into nets i mean catching a whale in a net is actually kind of hard if you're if you're trying to get small fish but catching a dolphin that's half the size of a human that you know that that goes into the net just fine, thanks.

And so they end up as bycatch. I remember estimating there are fewer than 10 vaquitas alive as of February 2022. Yes. I think last I saw, they were trying to do a captive breeding program, and I think that was quite a few years ago as well. So I'm not sure if the update is that, but they got one, and it did not succeed. So that was very depressing as well.

But yeah, the ocean has a hard life, because they are bycatch, like you said, but not only bycatch, but bycatch of poaching, for the most part, operations for a fish that's not even, they don't even want to catch the taquito, right?

Oh, that's so sad. Yeah. This sort of research that the trackers do help us determine that animals are being hunted and experiencing decrease in population towards extinction. Do you see that in the data or do you only get little blips? How much of the data do you get to look at? So the data is available for me to look at, but I don't get to look at it too much.

It's mostly anonymized not on purpose but just because of the bulk of it anonymized in the sense that i don't always know what animal it's from or what kind of animal it's from um sorry i would have to dig down for that so yeah i kind of just see blips okay um i'm hoping this is this is more amusing um what is a fish hackathon so fish hackathon i did that that again while I was in school. And that was, it still goes on, it's every year. And it's a hackathon. But it's all geared towards fish and marine conservation topics.

So I've been to a few hackathons, I guess, where they kind of just open and you do whatever you want and so the when i did this fish hackathon and i did it long beach uh they kind of gave us a list of desired projects that you know organizations non-profits and different companies had come up with for us to work on and so you kind of chose one and worked on that project so it was very interesting um i'd love to see more of that uh around embedded hardware because that's one thing i've noticed. I've been to that one and a different one that the San Diego Zoo, I think, put on. And they were all, I think, hackathons. If you only have a weekend, a lot of apps and web-based projects. Making things small is one of the hard things in engineering. So it would be hard to have a hackathon for that.

But just talking about the sensors. And I mean, the idea that the crested occlites fluoresce is really interesting. I saw on Nautilus Live, there was a question of, have you ever run Hercules, the ROV, with UV sensors? And the folks who were running it were like, no, we haven't. That would be neat. And I think there could be more hackathon ideas like that that are like, well, I have

this sensor. Can you use it? Yeah. Are you familiar with Nautilus Live? Yes, I am. Wonderful. So which is better, Nautilus Live or ZFrank? I don't know. Was it ZFrank? I don't know.

Was it Z Frank? So maybe now I'm going to have to look that up and figure out which one's better. What was the second one? Second one is impossible to describe. They're true facts about animals. The true facts about the nudibranchs. They're nature videos, sort of. But they're very silly. They're like little compilations and they're very silly. Compilations and they're

narrated by a guy who I think started out trying to sound like Morgan Freeman but then just ended up sounding like something else. There's a lot of jokes and it's definitely not safe for work. Well, I mean, the videos usually are.

Now that I think about it. Let's go with not safe for work. A couple more listener questions from Ben. What is the airspeed velocity of a tracker-laden swallow? I think the proper response is African or European, right? Exactly. Indrek would like to know, what are the best and worst uncommon animals to track and why? Do they have to be uncommon? Yeah, let's make that easier to answer. But I guess the easy answer for best is, of course, elephant seal.

Not only because I love them, but it's an interesting thing that they are also the perfect animal and this is opinion my opinion of course but they do have this interesting characteristic i think i mentioned before that they dive extremely deep right um and then they transit if they want to and then a lot of times they'll come straight back up and so if you imagine that looking at that profile in depth you basically have a an animal buoy that you can put instruments on and they always come back to the beach in the same location right and you can grab it off and get a full archive and so they are kind of like the perfect animal ocean scientist and that's a whole another area of research is using animals as instrument platforms for non-animal, right? Because you assume all that data is for behavior for animals, but a lot of

times the data can be for environment, for global warming research, for just atmospheric research. NOAA will spend a lot of money putting buoys out, and the buoys just go down and up again to sample the depth and temperature and salinity at different depths, right? And so you can imagine

elephant seals are perfect. Animal buoy, they can do that same job. And so, yeah, elephant seals are great. There's no worst. All animals are great. What's the hardest to deal? Bees. They're all great. Maybe fish.

Fish, yeah. You know, the pop-off tags, they're hard. The smallest ones are probably the hardest. Very small pop-off, the ones that sample for a long time and you don't hear from them. And if you never hear from it, did it break? Did it get eaten? Fish are very, very untrustworthy. Yes. So you mentioned transit.

And I came across that term with leatherback turtles, because transit means they go across the ocean, and they're not going to pelagic areas where they want to be in the middle of the ocean. What's pelagic mean for people who don't know what that means? The middle of the ocean. What's pelagic mean for people who don't know what that means? The middle of the ocean. Their food is almost always near a coastline. Maybe not, you know, 50 feet within a coastline, but close to a coastline. So when you go through the middle of the ocean, it's basically a desert.

And animals just go right through. They don't eat. They don't stop. Some of them don't even sleep. Where are the elephant seals going that they want to get there without sticking to a coastline where they can eat? Again, that I'm not a biologist, but from the things that I'm remembering is that there are seamounts, I don't know if you've heard that termed, but these are these mountains under the ocean, basically, that don't break the surface.

And so they can be really high biodiversity areas. And there's algae blooms that attract. So there are islands of life out there in the Pelagic area that just life congregates, right? And there's lots of things to eat. And that's one of the things that's really interesting, if you figure out where those are and maybe how those move over time and as the oceans change temperature, for example.

I'm sorry, I think leatherback turtles are better than elephant seals. I'm sorry I think leatherback turtles are better than elephant seals I'm sorry I don't mean to be argumentative but they're really cool have you ever pet a leatherback turtle? we asked him I have not

my wife has well leatherback I'm not sure but she worked at a turtle rehab for a while and that's another animal that comes back and will bring your sensor back eventually. That one doesn't do it every year. You loved animals before you did engineering, before you went to school for it. What advice do you have for people who are listening and thinking, I can have a career where I get to play with animals and do engineering?

What would you suggest? Just to seek out. And I think it's like any other niche field where you need to network and just talk to people who are also interested in it. That's kind of what I did. I know when I was in school and before school, even when I was thinking about going back to school, I kept kind of a bookmark tab of all the different companies that worked in wildlife and different people and opportunities and kind of just emailed them. It's funny because the job where I currently work, I cold called them. They have no, they're a fairly small company and they have no, you know, careers page or I think we do now actually, but they didn't at the time.

And I just cold called the support line and they forwarded me straight to the head of engineering and I chatted with him and within the first conversation he invited me up to intern and so I think at least this industry is very friendly to that kind of interaction because they tend to be very small companies that's pretty cool I would never even consider doing that. Yeah. That's very scary when I did it, but it worked out very well.

Yeah. And the company size is, you know, a strength and weakness because we're on the large size for this industry, right? There's a couple about our size and then there's a lot of shops

and a lot of diversity and different tags that make maybe specialized tags like those lizard tags that I had used where the company is maybe two or three people. and a lot of diversity in different tags that make maybe specialized tags, like those lizard tags that I had used, where the company is maybe two or three people. Because it's all very specialized.

I mean, you're currently focused on marine life, but there are totally different requirements for insects or land or air, and they all need to have their own sets of trackers yes exactly and the weight and size and there's even companies where it's like a handful of people and they very much specialize in like a video recording tag or something maybe much more specialized in audio or just different technology are there things that you were this was another listener question are there things that your trackers would be good for other than tracking animals?

I think they're pretty specialized, but I have known our tags to be used on ROVs. So I'm not sure if MBARI uses any, but I know some organizations do put them. Maybe less the ROVs, I should say gliders. Gliders are the ones that wander through the ocean, look like missiles, but usually are checking in with other things. Our tags are very good at rough location, with the horizontal movement, with the Argos or the light.

And so, yeah, I think they do have kind of a niche if you need it for a backup or a secondary localization technique. And they have an environmental sensor, so that makes sense too. Yeah, it's a complete sensor package. You can just slap it on things. Do you get an opportunity to still do field stuff or are you mostly stuck behind a desk these days? I am mostly stuck behind a desk these days? I am mostly stuck behind a desk these days for work. Very, very occasionally. I haven't done it yet through wildlife computers, but

engineers will be able to go to the field and take part in the testing, and that's important. They believe that it's important to have that interaction. I mostly live vicariously through my wife, who is a biologist, luckily. Do let us know when you finally pet some interesting animals, besides elephant seals.

So they don't count? No, no, elephant seals count. If I could pet one, I would be very excited. Terrified, but excited. But yeah, I think you, do you want me to send you the list? We didn't even get through it. I had Weta on there as well, because that's another thing that i would like to i don't want to pet a weta i didn't have giant isopods because now you can just go down to the aquarium here and just pet the giant isopods lemurs

ethan do you have any thoughts you'd like to leave us with uh either just that nature is all around us and i think one thing i've learned is you can always just go outside and there's always something if you look closely enough, some animals living their lives around us. That is true. Christopher got me a little tiny microscope camera for my iPhone and I found out there was

a really, really tiny snail living in a piece of moss. I would never have found it. Our guest has been Ethan Slattery, Embedded Software Engineer at Wildlife Computers. Thanks, Ethan. Thank you. Thank you to Christopher for producing and co-hosting. Thank you to our Patreon listener Slack group for browbeating Ethan into agreeing to do the podcast

and then asking a bunch of questions, most of them having to do with eating trackers. And of course, thank you for listening. You can always contact us at show at embedded.fm or hit the contact link on Embedded.fm. And now a quote to leave you with from The Genius of Birds by Jennifer Ackerman. A new study comparing the genomes of birds suggests that genetically speaking, the turkey is closer to its dinosaur ancestor than any other bird is. Its chromosomes have undergone fewer changes than other birds since the days of feathered dinosaurs. So why do they smell like tangerines? Again, the why is always hard. I think the assumption is, though, that it always comes back to being sexy for the mates, right? Sexual selection, especially these animals that have these crazy fluorescent bills, you know, that fall off.

Obviously, when they're not in the mating season and the big crest and the smell, it's all... Did you take a tangerine with you? And compare it? Yeah, to find out if you were sexy if you had a tangerine with you? And compare it? Yeah, to find out if you were sexy if you had a tangerine. I wonder if they've... So one of the people we were there with was doing a study where they built these little dummy auklets, which were very cute, very, very cute,

and put different sized crests on them, and then put them around the colony and saw how much interaction they got depending on crest size. And so they should do one with the smell, but...