Transcript from 377: Robot at the Park with Erin Kennedy, Elecia White, and Christopher White.

Welcome to Embedded. I am Elecia White, alongside Christopher White. Our guest this week is maker Erin Kennedy, also known as RobotGrrl.

CW (00:17):

Hi, Erin. Thanks for joining us.

EK (00:19):

Hello. Thanks for having me.

EW (00:21):

Could you tell us about yourself as if we met at lunch at Supercon?

EK (00:27):

Sure. So, hello. My name is Erin. Online my nickname is RobotGrrl. And the reason for that is because I really enjoy making robots. One of the areas that I enjoy making robots the most for is our environment.

EK (00:46):

That led me to create an initiative called Robot Missions, where one of our first robots is one that picks up plastic on the beach. And we've had all sorts of adventures with this robot.

EK (01:00):

I really enjoy designing the robots from the ground up. So the 3D design, the fabrication, the electronics, the firmware programming, and some software as well. I just really like robots.

EW (01:23):

I think that's admirable.

CW (01:26):

Who doesn't like robots?

EW (01:28):

Exactly.

CW (01:28):

Well, evil villains... -

EW (01:31):

- and evil robots, probably.

CW (01:36):

Yeah.

EW (01:36):

Are you ready for lightning round?

EK (01:36):

Sure.

CW (01:37):

Should learning be hard?

EK (01:39):

Yeah.

EW (01:41):

Complete one project or start a dozen?

EK (01:42):

Start four and iterate three times each.

CW (01:48):

Okay. Do you have a favorite motor?

EK (01:52):

Servo motor.

EW (01:54):

Which Sesame Street character best represents you?

EK (01:58):

Kermit the Frog.

CW (02:00):

What planet or moon would you visit if you could?

EK (02:04):

Saturn.

EW (02:06):

Do you have a favorite body of water?

EK (02:10):

The one on the Moon sounds cool, at one of the pools.

EW (02:14):

Okay.

CW (02:15):

Favorite fictional robot? Did we already ask that?

EW (02:19):

No.

CW (02:19):

Oh, okay. Favorite fictional robot.

EK (02:22):

WALL-E.

EW (02:24):

Do you have a tip everyone should know?

EK (02:26):

Remember to take breaks.

EW (02:30):

Okay. So we want to know more about your robots. You've been working on ones that try to clean up plastic from shores. Is that right? Or...is it from the ocean itself?

EK (02:44):

It's land-based. So terrestrial, from the shores. Yeah.

EW (02:49):

And when I go pick up trash at the beach, it's pretty tough. I mean, as a human, I can identify things that don't belong and things that do belong. And even then it's tough to figure out how to pick it up right out of the sand.

EK (03:04):

Yeah.

EW (03:04):

How do you get a robot to do that?

EK (03:06):

Well, the approach at first is very rudimentary. So sort of brute force, like use a shovel that's attached to a robot arm, and scoop it up, and dispense it into a hopper. So there, you kind of have a mechanical filter with say, a mesh on the shovel, so some sand falls through.

EK (03:30):

I guess it helps to give some context here that the plastic we're going for is sort of the smaller type as well. So, the methodology there would be that we would likely filter the sand and the small plastic afterwards as well. So that handles the picking up aspect.

EK (03:57):

And then there's also the aspect before that, which is detecting the plastic. So...detecting things used to be very difficult. Kind of still is. But luckily we've had a lot of advancements that make it easier to run these experiments in the field without too much expensive equipment.

EK (04:22):

So for example, we can train a data set based off of images that we've scraped from online or images that we've collected ourselves of various debris on shorelines, then use that to create a model which is used for either image classification or object recognition.

EK (04:48):

That model can be run on a computer, like a Raspberry Pi, and then you have that working in the field. And that's how it can detect things. So what we ended up doing was actually an image classification approach, which ended up being the wrong approach. Because object recognition would have been a bit better.

EK (05:15):

So that way we wouldn't have to have segmented the image 9 times to sort of get its location within the field of view of the camera. But yeah, basically it does work.

EK (05:29):

We were able to detect basically debris versus non-debris. Which was an interesting thing, because one of the mistakes we made at the start was not training it on natural debris, such as the sand that's already there, or twigs.

EK (05:47):

So then it was just detecting everything as garbage, and that was incorrect. So it was a cool learning experience to be like, "Oh, it also has to detect what has to stay so that way we can filter that." But, yeah.

EW (06:02):

Your data sets are really important, and people do forget that the uninteresting case is just as important, if not more important, than the interesting case that you're looking for. So did you use TensorFlow, and Python, and all of that to do this?

EK (06:19):

Yup. TensorFlow.

EW (06:21):

And, you said image classification, and then segmentation. What other algorithms are you looking at? I guess you said something that made me think of YOLO but how did you choose this algorithm versus others?

EK (06:40):

We didn't choose it very well. So the two types are image classification and object detection. Object detection is just like what you mentioned, YOLO, where it detects in the frame the sort of bounding box of what it has detected.

EK (06:59):

And then from there you can get the width, height, x, and y position, whereas with image classification, it's classifying the whole image. So, what we ended up doing is making 9 small images from that one big image. So that way we could get the actual location sort of within the camera.

EK (07:27):

So what went into that decision was, we were new to TensorFlow, and we saw, "Image classification, here's a way you can try to do it." And I was like, "Wow, this is very cool. Let's try it." And then only later on, did I learn about a better way of doing it.

EW (07:49):

But this is from the software, as we've been discussing, but also all the way to the mechanical parts. Isn't it kind of tough to run a robot in the sand? Are you in sand, or are you in silt?

EK (08:03):

We're in sand.

EW (08:04):

Okay.

EK (08:04):

Yes. It's incredibly challenging for a robot to navigate in sand. There's also different types of sand. It's sort of deceiving how challenging it is.

EW (08:19):

What are the hard parts?

EK (08:20):

Oh, boy. So the hard part is wheel geometry, wheel material, turning, the type of sand, the granularity of the sand. If it's wet sand that goes very deep. Or if it's wet sand and then dry sand underneath. There's payload weight, and motor torque, saving the easiest for last.

EW (08:51):

Are you looking in a specific area for the debris, like the strand line,...where the water is ideally receding from the strand line, and so you get the crusty sand, but you also get a bunch of debris in the strand line? Or do you have a different area?

EK (09:14):

Yup. That's one of the locations that is one of the easiest places to look for debris, because if it has been in the water, that's where it will wash up. And we see that, especially when there's been a recent overflow, say from a heavy rainfall event, so then you'll find a bunch of plastic around that line.

EK (09:40):

The funny part though, is that on beaches, you'll often find debris in other places too...because of human behavior. So there's two factors when we're talking about -

EW (09:53):

Pick up your trash, people! Sorry.

EK (09:56):

Yeah. Yeah, exactly. There's these two factors when it comes to terrestrial plastic debris, it's the stuff that washes up, and then human behaviors. Kind of disappointing, but yeah.

EW (10:13):

Okay. So...your robots, is the goal for them to wander around by themselves? Do they have a localization, navigation, sort of thing? Or are you overseeing them more carefully?

EK (10:31):

Oh, yeah. Yeah. That's definitely the goal, have them wander around by themselves. So what we accomplished for autonomous navigation ended up being using markers in the field.

EK (10:48):

So we placed these augmented reality markers on the beach and then had the robot detect those and navigate within that zone. For our experiment purposes, the zone was about 1.5 by 1.5 meters. So when I first started out on the autonomous navigation aspect of the robot, I started out using GPS.

EK (11:19):

However, at that time, it was just, let's say regular GPS, where you're receiving the NMEA strings and then getting the location from that. The problem is, is that our zone size is small, as I just described. And the error for GPS, using that method of GPS, is larger than that.

EK (11:42):

So what would happen is the robot would sort of just end up moving its zone throughout the day when it was using the GPS method. So yeah, it sometimes drifts so much that it was close to the water.

EK (12:06):

And one time we had one,...when I was working on another replicant of our robot and letting the other one roam free, so focused that then the other one just roamed into the water. It was so close to the battery. Oh my gosh. It would have been an interesting sight.

EK (12:29):

But yeah, so maybe there's another question that can build on this question.

EW (12:36):

Oh yeah. I mean, part of me wants to go towards, "Well, of course they're going to get wet." In the ocean, the waves don't tell you when they're coming. But the other side of that is, so GPS, yes, it's not exactly precise. And as a robot that's moving, it can be a little confusing as to the timing.

EW (13:02):

And then you also have the timing of the images, and whether or not you're scooping, and do you slow down to scoop? So there is a lot of localization ... gotchas. Do you use a formal algorithm for this? How did you learn about robotics enough to drive the robot like this?

EK (13:22):

So for the first question, yeah, it was sort of touching on the timing aspect, right? So there's multiple systems within this robot. There's the plastic detection, there's the actions of scooping, there's the navigation, and then there's also the driving. So the timing, it would depend on which mode we were using.

EK (13:52):

If we were using the mode where it's time for the robot to clean, then how it would work is, it would just be traveling in a straight line. So here's the flow. So drive forward, take a picture, classify it.

EK (14:15):

If there's debris, scoop. If there's no debris, then move forward about 5 to 10 centimeters. The way it worked, saying that, that's great theory. It sounds fine on paper. It sounds fine in the code, but -

EW (14:33):

What could possibly go wrong with that?

EK (14:36):

So how it worked in reality is that the Raspberry Pi takes a long time to process the image. So it would be this robot that would inch forward. And then it stops, and it's waiting. And you're looking at this robot, and you can't really tell it's thinking, right?

EK (14:58):

So you just think, "Oh, it's just now stopped." But then it will do something if it detects plastic. And then it will drive forward again, and then it will stop.

CW (15:10):

You need some sort of little thing to put on top that spins around really fast when it's thinking hard.

EW (15:15):

Or just a light. The thinking light.

EK (15:18):

The thinking light.

EW (15:18):

Like a light bulb.

EK (15:19):

Yeah, exactly.

CW (15:19):

Yeah, well. Okay. Alright, fine. Yes. That's what I was probably thinking of.

EK (15:28):

But, this was actually just a really great result of the experiment, because we wouldn't have known this unless we tried. And that was a few years ago now. There's so many more advances now. There's the NVIDIA Jetson Nano 2GB. So there's these boards like this now that make it a lot faster to do that processing.

EK (15:56):

We're also seeing more ways to do connectivity. So perhaps processing on the cloud might be more of a reasonable approach. And, finally on top of that, there's now even RTK GPS breakout boards...That would have been nice. Yeah.

EW (16:16):

I use the Jetson TX2, which is the big brother to the Jetson Nano. The older big brother.

EK (16:23):

Nice.

EW (16:23):

And yes, it is much faster than Raspberry Pi. Are you still working on the shore cleaning robot?

EK (16:32):

A little bit here and there.

EW (16:36):

What's happening with it?...Is it mostly a research project, or...? I mean, I want you to say people are building it, and you'll see them at shores wherever you go. Soon it will be larger than iRobot. But what is happening?

EK (16:57):

So what's happening right now is we've had a few people replicate it. So they 3D printed all of their pieces. Oh, yeah. We probably should have mentioned that the whole robot can be 3D printed.

EK (17:12):

So there's been some people who have replicated it. And where we want to see it go would be for it to be a kit that really involved organizations, groups. Or people would be able to take on this project to be able to deploy it at their shorelines.

EK (17:35):

The stumbling block, though, is that this has been a really key learning moment for me throughout this project. Because it's been 23 iterations to get to the sort of version 1.0 state...Each system within the robot is essentially like its own product.

EK (18:00):

So for example, there's 14 components within the Bowie robot itself that would essentially be their own kit to make this entire robot kit. And....it's a lot to tackle. But yeah.

EK (18:25):

So the vision would be to allow these groups to replicate the robot. Where we're at now is kind of stuck a little bit on the documentation, getting everything together, and doing that times 14 times.

EW (18:46):

It is a big project, and it's a large stopping block for a lot of people. How long have you been working on the shore cleaning robot?

EK (18:55):

Been working on it since 2016 and sort of formally incorporated it in 2018-ish.

EW (19:09):

And...when I look at robotmissions.org, there's a season finale video. What is that from?

EK (19:20):

[Ah], yes, the season finale video. That was from our Mission Pilot, which we did in collaboration with the city of Ottawa, which is located in Ottawa, Ontario, Canada. And what we did was we ran a Mission Pilot of these robots being deployed on the beach, and we would give our results to the city and hear their feedback.

EK (19:47):

And we really worked on developing the robot to the state where it could pick up garbage, sort of, and it could detect the garbage for sure. And we invited the community to be part of this through field tests.

EK (20:08):

Field tests is when we would try the latest features that we've been developing on the robot and essentially hand over the controller to kids, people in the community, to try it.

EK (20:23):

And that is the most spectacular way to sort of fail and learn from that, gather observations, and also get feedback from people in the process of, "If these robots were to be adopted by communities or by cities, maybe it would need this sort of functionality to be easier to understand, in order for it to be more successful."

EK (20:49):

So, the community really enjoyed it. And it was just a very community-oriented initiative. We also did a lot of tech dev. One of my favorite moments of it is when two girls showed up at the field test. Two young girls showed up at the field test, and they didn't know much about robots prior to this.

EK (21:24):

And so they started to play with Bowie the Robot, and they were super engaged with it. Next thing you know, the next field test we had, I think their mom came back and said, "Oh yeah, they actually got a Lego robotics kit. And they're building their own version of Bowie now too."

EK (21:49):

And it's like, "Wow, that is so cool." And it was one of the most surprising things of just realizing that if we bring robots to a place where it's not really expected, then you reach so many more people to get inspired to consider robotics.

EK (22:12):

It's...out of the classroom, it's at the park, and you wouldn't expect a robot to be at the park. And it really catches the imagination, I guess.

CW (22:28):

Was that one of your goals with the project partially, is outreach, and kind of making people aware of robotics in a different way?

EK (22:37):

Yeah.

EW (22:39):

How did you learn to do this? I mean, I know you're not as old as we are. I know that. But -

CW (22:50):

I'm sorry, I'm just offended over here.

EW (22:52):

Are you offended? I'm pretty sure. I remember when she was 16, and we weren't.

CW (22:58):

Okay.

EW (23:01):

How did you learn about these things? I mean, navigation isn't easy. AI isn't easy. Robotics isn't easy.

CW (23:11):

Beaches aren't easy.

EW (23:12):

Yeah. How did you build up this knowledge?

EK (23:20):

How I built up the knowledge was through lots of online learning, lots of various robot projects, online communities, and mentors who are so gracious to lend their time and expertise really. And yeah, just trying a lot, failing a lot.

EW (23:47):

And...this project and many of your others are on the maker side of the spectrum between maker and product engineering. But you also do some professional engineering. Can you talk about that at all?

EK (24:06):

Yeah, just a small correction. Can't really say I do professional engineering. That's actually a designation here in Canada and -

CW (24:15):

Yeah.

EK (24:16):

- I do not have that.

EW (24:17):

Oh, right. The P.E. thing. Yes.

CW (24:19):

Yeah, yeah, yeah.

EK (24:19):

Yeah.

EW (24:19):

Okay. So you do professional software?

CW (24:23):

Professional computering.

EW (24:25):

Computering.

EK (24:28):

Usually when -

CW (24:29):

Computering for hire. I don't know how to legally -

EW (24:30):

Making money. How do you make money?

EK (24:36):

So how I currently make money is, I'm really lucky to have a part-time job working at the prototyping lab at an innovation center called Bayview Yards in Ottawa. Some people may know it also as Invest Ottawa. So I work in there, where we have a workshop with a digital side.

EK (24:59):

So we have a bunch of 3D printers, a water jet, big 3D printers, CNCS. And I get to help early stage entrepreneurs sort of like myself work on their hardware prototypes. And my area that I focus on is on the electronics, and firmware, and software side.

EK (25:26):

However, depending on the demand at the time, it can venture into the 3D design side as well. And so yeah, I mentioned that's part time. So then, I don't make money off of Robot Missions just yet.

EW (25:44):

You list your education on your website as unconventional. What does that mean?

EK (25:51):

So unconventional, meaning it hasn't exactly followed the same path that many have done in order to gain their expertise, their well-earned expertise, in these fields.

EW (26:13):

So you didn't go to university.

EK (26:18):

Well, luckily now I can say I have been to International Space University, but it's not a formal, 4 year degree, let's say.

EW (26:30):

Well, what is International Space University? That sounds like fun.

EK (26:34):

Yeah. So International Space University is a educational institution. It's headquartered in Strasbourg, France, and they have a few programs. So the one that I did was called SSP. So this takes place every summer.

EK (26:57):

And through this program, you get to learn everything about space, and what makes it interesting is that it is international. So they bring together a cohort of about 100 people each year who are either interested in getting into the aerospace sector a little bit more, or sort of into it already and looking to build on their skills even more.

EK (27:25):

So it's this gathering of people who have knowledge from all sorts of different areas, and then we're combined together to learn more about space. And I was so lucky to be able to go there, much thanks to the European Space Agency sponsorships and crowdfunding.

EW (27:55):

You have worked on other space projects for yourself. What...other projects...have you worked on?

EK (28:05):

Some of the space projects? I guess my first dabbling in it was this one called Rapidly Deployable Automation System. And the premise for that was that it was a cube set that would unfold to become a swarm robot. And so I built that, and it was pretty cool.

EK (28:31):

After that, in terms of space, there was one where I advised. It was about a Saturn wheel. If we were to go to Saturn, how would we navigate in order to get the most data.

EW (28:45):

How would we?

EK (28:45):

Well, it would be...more of a mechanical clockwork system. So unfortunately, I was just the advisor for it, so I didn't really get to be hands-on. But the team ended up creating something similar to a Hoberman sphere, where it can sort of expand and contract.

EK (29:07):

And by doing different sequences of that movement, they would sort of have some sort of directionality, depending on how it was constructed. And that was the idea. It was very imaginative and could require additional research for making it more realistic.

EK (29:31):

So, let's see. Other space projects after that, got really into building towers with discrete units as sort of the premise of our Bowie bots. Then got really into wireless power transmission, but it didn't really go anywhere. And now that leads me to my latest space project, which is the Mars Wind Tumbler.

EK (29:58):

And this one might be something that actually amounts to something. So there's this project. Its idea is to use wind as the primary propulsion method for making a rover move on Mars.

EK (30:19):

And we proposed this to the Transatlantic mission for the Mars Desert Research Station in Utah...as part of an analog astronaut sort of mission. And they accepted it. So now we get to build it, because we know it will be deployed on Mars, well, on Earth, but yeah. So, yeah, that's the latest space project.

EW (30:50):

Are you familiar with the NIAC grants? I guess you're Canadian. So maybe not, although -

CW (30:58):

I don't know how that would work, if at all.

EW (30:59):

I do think they do international.

CW (31:02):

Okay.

EW (31:02):

...Their deadline is in July. And so I'm back to looking at what crazy space idea I might possibly have that I could get people to pay me to look at.

EK (31:16):

Ooh, that sounds cool.

EW (31:18):

Isn't it? Yeah, we talked to Derleth, Jason Derleth, one of the people who judges the entries, and it was pretty neat to hear about some of the slightly crazy things people want to do.

EW (31:35):

But you never know if crazy means crazy good or just not physically possible until you try it sometimes. Except for perpetual motion. That is not possible.

EK (31:53):

Yeah. Whenever I hear those things, why is it always that that comes up first? But, yeah, that sounds cool. I'll have to check it out more.

EW (32:05):

As I looked through your website, I saw your autumn 2020 project roster, which listed a whole bunch of projects, more than anyone could possibly have time for. But you also list learning goals for each project. Is that how you choose projects that you're going to work on? Or do you look more for the application?

EK (32:27):

Sometimes. So from what I've seen, it's important to still have projects where you can let your just play or just follow the next interesting thing that appeared in your project. There's not much applications for those usually, unless it happens to be a pleasant surprise.

EK (32:58):

So how I usually try to go about thinking about which projects to take on nowadays is dividing it by things that could have an application to a problem in the world. And the timeline for that could be between now and 10 years.

EK (33:25):

Then the next stage projects I consider is a project that can have an application to solving a problem in the world and also have a market. So that's more of the projects that could be entrepreneurial, because those ones specifically are constrained by, there has to be a market need for one of these, whatever, things, projects.

EK (33:58):

And then the last category is just frivolous projects, imaginative things that help keep my spirit alive.

EW (34:11):

How do you decide when a project is done?

EK (34:14):

Yeah. I don't know if I have a satisfactory answer for this, because I probably should have given up on robot missions a very long time ago.

EK (34:27):

Maybe the deciding factor for when a project is done is when you're no longer having fun with exploring it in different dimensions, whether that could be the application side, the research side, the market side.

EK (34:43):

If none of those are fun, then it might be time to pick up a different project or just not do projects for awhile and sort of gather information about the world. Yeah, yeah. Not a satisfactory answer at all.

CW (35:00):

No, that's...really good. And it's one that I think I've accidentally done without consciously thinking about it.

CW (35:09):

Just deciding when you're not interested in something, whether it's deciding whether you're not interested because you've been spending too much time on it, versus it's not going to pan out. That can be a difficult kind of thing to tease out, especially when you're already tired.

EW (35:24):

Well, and I've had this revelation that hobby and career should be separate for me. I know some people can do both, which, it sounds like you have a good crossover, but for me, every time somebody wants to pay me for my hobby, I start not wanting to do the hobby.

EW (35:46):

And so I keep moving further and further away from what people could possibly pay me for it, which is how I got to origami snails, because who would buy those?

CW (35:56):

I feel like that was a deliberate decision. Like -

EW (35:58):

It was. It was.

CW (35:58):

"This is the least marketable thing I can do."

EW (36:01):

Exactly. But yeah, you have to balance what you do because you want it to get done, what you do because it feeds some part of your inner creativity and happiness, and what you need to do to make money.

EW (36:20):

And those things don't all have to be the same. And I think it's a good realization that you do need to do all of them, but they don't all have to be the same thing.

CW (36:34):

Along those lines, I want to go back to your frivolous, wow. Your frivolous projects. There we go. I want to go back to your frivo, I can't do it.

EW (36:45):

He doesn't cut his own stuttering, by the way.

CW (36:46):

I want to go back to your silly projects, and hear more about those, because those are often things that lead accidentally to maybe non-silly things in the future. But what do you like to do for silly projects?

EK (37:00):

Well, to be honest, I only really got back to silly projects recently. But the most recent one is this one called fog voxel, and it's a volumetric pixel made out of fog. And it's illuminated with RGB LEDs scattering the fog.

CW (37:23):

You mean like a cloud chamber? How does this work?

EW (37:30):

...Okay. Here's my image of it, which is probably totally wrong, but is it a 6x6 set of clear boxes that you can pump smoke or water vapor into and suck them out as necessary?

EW (37:48):

And they have LEDs, and you can make something opaque, or you can make it clear, or you can make it blue, or green, and okay. Sorry. What are you actually building? Please don't let my ideas influence you.

EK (38:02):

That's so cool. Yeah, that's pretty much it. So yeah, that's really cool. Basically it's this ice cube tray. And there's water in the ice cube trays, and then lots of hot glue to hold some things in place, such as these piezo discs, where if you oscillate, then they'll change the water into a vapor. And that sort of expels the fog.

EK (38:38):

And it's a grid of 3x3, and the RGB LEDs point up, and it sort of illuminates the fog plume that's sent by these discs. And I've just been having fun with this. And there's no application. Well, actually, so I've been thinking of many applications.

EK (39:04):

There's the entrepreneurial side of my brain, but no real big applications for this. And it's just been fun. And also if the disks run out of water, they sound like those screaming plants in Harry Potter, and it's just kind of adorable.

CW (39:25):

Okay. At the risk of getting into too much detail, we have a Halloween fog machine thing.

EW (39:30):

I was thinking that, that it was probably the same technology.

CW (39:34):

And that was expensive back when we got it, which was probably a long time ago.

EW (39:40):

'90 something.

CW (39:40):

So tell me about these piezo things. Are these just off-the-shelf little parts?...I've tried to make -

EW (39:51):

Speakers?

CW (39:53):

I tried to make a drum...triggers out of little piezo, I don't know how to say that. The little discs. Are these the same things? Are they just little round metal discs?

EK (40:06):

Yep.

CW (40:07):

Wow.

EK (40:07):

I believe they might have some holes in them that are very tiny, but if you've ever used a diffuser with essential oils -

CW (40:19):

Okay.

EK (40:19):

- it's that piece that's in the bottom. And then, yeah,...so I'm sort of saying it nonchalantly here, but it totally blew my mind when I learned about this thing. You can turn water into gas just by vibrating a disc? What? That is so cool. So I've been kind of obsessed with these.

EW (40:45):

You can get 5 of them on Amazon for about 10 bucks.

CW (40:50):

Okay.

EW (40:51):

I mean, it ranges between...3 and 12, but yeah. So you can get a lot of them, and -

CW (40:59):

Do they drive themselves, or do you have to put a signal on them?

EW (41:02):

2 wires, red and black. So I assume you drive them.

CW (41:08):

That was a question for her.

EW (41:09):

Oh. Oh, yes. Erin, as somebody who's actually touched them, how do you drive them? PWM, or just on, off?

EK (41:19):

So you have to boost up the voltage a bit and then use a 555 to -

CW (41:28):

Oh, okay.

EK (41:28):

- oscillate it at around 113 kilohertz.

CW (41:32):

Okay. So super ultrasonic.

EK (41:34):

Yep. Yep.

EW (41:36):

What voltage do you put into them?

EK (41:40):

Well -

CW (41:41):

They're speakers.

EK (41:42):

5 volts from the Arduino. Then you have your boost-up circuit. So...I'm using little modules for this one. I think it boosts it up more than 12 volts at least.

EW (41:59):

Sorry. I have a motor driver that would boost voltage as well as current...Okay, I'll stop surfing Amazon for -

CW (42:07):

Fog machines.

EW (42:09):

Fog machines. That does sound like fun. I want to get, you said, an ice cube tray.

EK (42:17):

Yeah.

EW (42:17):

Could you do it in a baking sheet pan where they're separated, so they aren't cups, but...over here it, has more fog than over there, depending on, never mind. I'm using my hands a lot. It makes a lot more sense if you can see me moving them around.

EK (42:38):

I understand. Yeah. So trays that way, the water is sort of continuous, not individualized. But this is sort of where the entrepreneurial aspect comes in, because imagine if we could put different liquids in each of these individual containers or ice cube -

EW (42:58):

Shots.

EK (42:58):

- container things.

CW (43:00):

[Hmm].

EW (43:00):

Yeah. Sorry.

EK (43:00):

So -

EW (43:01):

Yeah.

EK (43:02):

So it's interesting, because I think one of the things that we learned in the pandemic is how our olfactory senses can be used as an indicator for some diagnoses. However, we don't really have a controlled method of doing that just yet.

EK (43:22):

And if it is controlled, like say baking an apple pie, well, that can also be subjective, because you know that you're baking an apple pie. So are you smelling it for real, or are you smelling a memory?

EK (43:36):

So, it's fascinating to see, even just at a superficial search level, how many diseases, where...if you had data about if your smell is decreasing, then it could help a little bit to at least give an indication.

EK (44:00):

So all that is to say is who knows, maybe you'll see these fog voxels at a doctor's office, and you'll have to go in and do a yearly Smell-o-gram test.

EW (44:16):

I mean, we have eye tests, we have hearing tests, we have touch tests. It's kind of weird that we don't have a smell test. And I do admit that as I ground coffee, most mornings, I would take a big whiff of it to make sure I could smell it. I know that that was not really diagnostic, but it always reassured me a little bit.

EK (44:39):

Nice.

EW (44:40):

Erin, you've always been a part of the maker community. As I said, I kind of knew about you from Maker Faires years ago, and remembering roboticist, and 16-year-old. How did you get into the maker community, and how did you stay a part of it?

EK (44:59):

So getting into the maker community, it's kind of an evolving process, always still is evolving. I was lucky to go to Maker Faire New York and sort of meet people that I've hung out with online there in real life. And it was super cool.

EK (45:26):

And then from there, just kept sort of building, like going to other Maker Faires, going to RoboGames. I was lucky to be an artist-in-residence at Evil Mad Science [Evil Mad Scientist Laboratories]. And I learned so much there.

EK (45:44):

And it's fascinating now how almost every week I'm thinking about things that I've learned there that I can apply now. Yeah, it's just been an evolving process of learning from everyone who has been just so nice to share their experience. And I try to share it back.

EK (46:07):

I've tried to do some open source hardware. But as you heard with the Bowie Robot Kit, 14 products within a product, it's sometimes challenging, but try to share back the process of what I'm working on and progress.

EK (46:27):

Yeah,...I don't know where to go with it, other than just to say I've been super grateful for the maker community for involving me in it. And, yeah.

EW (46:41):

How do you join the maker community?

CW (46:47):

Well, there's an application you fill out.

EW (46:49):

Well, that's the thing. There isn't. And I know that I'm sort of considered part of it, but I don't feel like I'm part of it, because I don't really make anything. I mean, I made Ty as a robot, but that wasn't makery. I don't know why. How do you become a maker?

CW (47:11):

Elecia has a weird idea of what the maker community is. Can you describe what the maker community is? So we can put this to rest?

EK (47:19):

Wow. Gosh, that's so unbelievable to hear, because you're one the key people to the maker community. You're literally bringing everyone together and giving us such a great platform to share the stories behind what we're making with each other.

EK (47:39):

And it's easy for everyone to just say, "Oh, did you listen to this episode of these robots or this person?" And it's so key. So that's surprising to hear, and hopefully you consider yourself a part of it.

EW (47:59):

You're so kind.

EK (48:04):

So, yeah, in order to become a part of the community, it's mainly just sharing what you're making. And then, say for example, if it's on a website like Twitter, you can share what you're making. And if you're just trying to get into it, tag some people like Make, Adafruit -

EW (48:32):

Hackaday.

EK (48:32):

- Evil Mad Science to Hackaday. Yeah, there you go. And then if they retweet it, even if you say, "I'm new to the maker community," they'll probably retweet it in that way. People can follow you, because they want to see what you're making too.

EK (48:51):

And, yeah, that's basically it. It's a very easy way to get into the maker community. The other part is, try to contribute back as well. And I think I should do a better job of that to be honest, but yeah. So share what you're making and contribute back. That's how you join.

EW (49:18):

And it doesn't have to be share every line of code, or discuss everything with everybody. Sometimes just saying, "I'm going to try to build this," is enough to let somebody else look and say, "Oh, that's a good idea. I can try to build something like that too."

EW (49:39):

Or it gives permission for somebody else to say, "Oh yeah, I don't know how to do that, but I'm going to try. Might as well." And that sort of being part of the maker community, I like very, very much.

EW (49:54):

And not necessarily about being an inspiration, but about giving permission. Not that anybody really needs it, but people sometimes feel like they do.

EK (50:07):

[Affirmative].

EW (50:07):

And I think you do that very well.

EK (50:10):

Thank you.

EW (50:10):

...Speaking about Hackaday, you were involved with the Hackaday Prize. Which year, and what did you work on?

EK (50:23):

I've done all of the Hackaday Prizes except for 2017, '18, maybe '19. But anyway, the most recent one, Hackaday Prize 2020, they had the Dream Teams initiative. And this is where Hackaday sort of paired groups of people together to work on impactful projects as a part of the prize.

EK (50:54):

And you even got a stipend for it. So the stipend part blew my mind because I was like, "Oh my gosh, I'm being paid to work on an impactful project and stuff that I'm good at? What? This is unbelievable." It's sort of the first time this has ever happened.

EK (51:14):

So, the group I was paired with was someone from Nigeria named Toby, and someone from Venezuela named Leo, and we worked together to develop a ropeless fishing gear robot.

EK (51:31):

And this was to reduce ghost gear in the ocean by being a buoy that goes from underneath the water with the traps to the surface at a certain amount of time. And we collaborated on this over 3 months, over 3 continents, completely remotely.

EK (51:50):

And it was awesome. It's so cool, because they also pair you with people at Supplyframe DesignLab, and I got to learn so much from them there too,...and got to learn from all the teammates too. It was just a phenomenal experience. And it looks like it will happen again this year.

EK (52:15):

So, yeah, the robot did work in the end. And now, hopefully this year, we'll get to make it even better and more professional, so that way it sets it up for success for the next steps, and get user feedback.

EW (52:34):

Neat. Ghost gear is a pretty serious problem, because animals of all kinds get trapped on the lines and in the traps. It would be better if that wasn't such a thing.

EW (52:48):

Alright, well, I don't know where I was going with that, but I do know we need to get back to our regularly scheduled days. I have really enjoyed talking with you. Do you have any thoughts you'd like to leave us with?

EK (53:03):

The thought that I'd like to leave with is just saying thanks for collecting all these stories, sharing them, and playing a big role in the maker community. So, yeah. Thanks, Elecia and Christopher.

EW (53:22):

Thank you. Our guest has been Erin Kennedy, known as RobotGrrl, and founder of Robot Missions. Thank you to Christopher for producing and co-hosting. Thank you to our Patreons for Erin's microphone. And thank you for listening. You can always contact us at show@embedded.fm, or hit the contact link on embedded.fm.

EW (53:45):

You might be surprised, but in this next week, there will be a bonus podcast that lets you know what's going on with myself and Christopher.

CW (53:56):

Wait a minute, that sounds ominous.

EW (53:57):

That does sound ominous, doesn't it? So, a quote -

CW (54:03):

About projects. It's projects.

EW (54:03):

Yeah, it's good stuff. [Music begins playing] It's all good stuff. A quote to leave you with, from Warsan Shire. "At the end of the day, it isn't where I come from. Maybe home is somewhere I'm going and never have been before."