In this episode, Audrow Nash speaks to Helen Greiner, who is the CEO at Tertill; Tertill makes a small solar powered weeding robot for vegetable gardens. The conversation begins with an overview of Helen's previous robotics experience, including at as a student at MIT, co-founder at iRobot, founder and CEO at CyPhyWorks, and in advising government research in robotics, AI, and machine learning. From there, Helen explains the design of the Tertill robot, how it works, and her high hopes for this simple robot: to help reduce the environmental impact of the agriculture industry by helping people to grow their own food. In the last part of the conversation, Helen speaks broadly about her experience in robotics startups, the robotics industry, and the future of robotics.

Start

Introducing Helen

Roomba’s backstory

CyPhyWorks

Helping special forces with your robots

Advising robotics and AI research investments

Introducing Tertill

Lessons learned from the Roomba

Challenges in making Tertill

Why make a weeding robot?

On gardening

Helen’s path

Matching the market and technology

Where and how to get information on opportunities

Advice for starting a robotics company

Different types of investment

B-to-B or B-to-C (Business to business or business to consumer) markets

Robotics consumer space

Future of robotics

Importance of robotics in the big picture

This is a conversation with Helen Greiner, who is the CEO of Tertill. Tertill makes a solar powered robot wieder for home vegetable gardens. And you see this Tertill robot, you might think of the robot vacuum cleaner the Roomba by iRobot. This comparison makes sense. And funny enough, they were both made by some of the same people, Helen included. What I think is beautiful about the Roomba, and this robot the Tertill is that it is so relatively simple, yet provides real value. Vacuuming under the bed is a pain. And I don't have much experience with weeding, but weeding sounds awful. And both of these products are for the consumer market, meaning that they have to be inexpensive enough that many people can afford them. The iRobot Roomba was one of the first successful consumer robots, and it grew the consumer robotics space, which benefited all of us in robotics, from making investment easier to come by to parts being cheaper to showing people that robots aren't all Terminator robots. As Helen says before this interview, and has been said before on this podcast, a rising tide lifts all boats. I look forward to seeing how the Tertill does in the future. And if I start a garden, I'll probably get one. I'm Audrow Nash. This is the sense think act podcast. If you like this interview, please consider subscribing. A big thank you to our founding sponsor open robotics. And now, here's my interview with Helen Greiner. Would you introduce yourself? Hi, I'm

Helen gainer. And I'm a robot enthusiast.

Now, we're gonna be talking mostly about Tertill. But I'd love to start, like for context to talk about your whole path in robotics.

You want to just tell me like, I mean, maybe at a high level from grad school, and then you were involved with iRobot?

Well, it started back when I was 11. And I saw Star Wars and fell in love with auto detail and odd to think, you know, he wasn't just a machine, right? He was a character. He had an agenda. He saved the universe. He had a personality, a lot of expression with his bleeps and bloops. And, you know, since that time, I've always wanted to make things that are more than machines. And so I went to MIT to learn how. And it turns out, I learned a lot of great engineering there, but they didn't really know how to build robots for the real world. And so right after grad school, I was in grad school at MIT, in computer science, although my undergraduates mechanical and accepted to the Ph. D. program, but instead decided to cofound iRobot corporation with vide books and Cullen angle.

And then, so you will What did you do at iRobot? So you were designing the robot I believe. Original vacuum cleaner robots.

Now, way back when way back when, when the company first found it. All of us myself lied and con, we're actually designing the robots and building them and fabricating them and everything about the robots. But though, you know, we started iRobot in 1990. It wasn't until, like, 98, we started seriously working on the Roomba and not until 2002. We put it on the market. So there's a lot of history before that. Oh, we are, you know, the longest overnight success. People hadn't heard about us until the bomba. But you know, there's a whole lot of robots that we were building. Before eventually military

robots for so many different industries. We did a robot for oil and gas industry doing downhole inspections, plug patch perforate stuff. We had robots that large cleaning robots, we had robots that were in a museum that was robots that we sold to research labs. We get a toy robot, a baby doll robot with Hasbro. Just so many different robots. But it wasn't until we put robots on the market under our brand that we were you know that we had any success.

Interesting. So it was was it kind of like you were contracts Acting around like people would want robots or you'd see it opportunities.

Yeah, it was it was kind of like any robot for money usually would get the large company to bring the dollars. And you know, we, I, we believe the marketing and the the market savvy, and we were doing the technology and the creativity and the innovation and the robots. And you know, it sounds good on paper. But what usually happened was the industries work at glacial paces. And the person who was advocate either move to another position or another company, and then you almost had to start the sales pitch, again, from the or from the beginning. Yeah. So I don't think we had enough top level buy in at the companies to make it successful. And in some cases, the robot technology was competing with the status quo of how the the existing product lines. So you know, a larger company wouldn't have that much incentive to, you know, push it forward.

Interesting. And then how did it how did you go towards the Roomba with us

like? Well, let's see, our first employee at iRobot was Joe Jones, Joe had been thinking about some of the things while he was at MIT, and that Maxwell was wide Brooks's behavioral control algorithms, because, you know, we really didn't have a lot of resources to put on these smaller robots. So we couldn't have, you know, the kinds of processing and the kinds of capabilities that some of the robots have today, even the Roomba. So, Joe wrote a white paper in I think, you know, maybe 9097, or 98, saying, you know, hey, we're building these large cleaning lots, we're building all these larger robots. But really, we should do the simplest possible thing first, and he spelled out, small, low bought five point about $200. Needs to be fully automated use the simplest possible senses rather than, you know, the most capable senses. And, you know, get something out there pretty quickly. And we, you know, it sounded like a good idea to everybody. So it was funded with very small amounts of money, because we didn't have a lot of money. And, you know, it kept building prototypes of it, Joe and the team, and they, you know, got, you know, we could never see a reason why it wouldn't work. And we fully expected when we started it to see a reason why there's one component missing, like, it's not a product. Yep. There was another robot vacuum. On the market. We didn't know that much about it at the time, I left on that Sandborn, but it was many 1000s of dollars. Us, it doesn't sound very high tech today, but use Sona technology, but it made it very expensive. And I think that when a new concept is new, it has to be reasonably inexpensive for people to take that jump. So Moonves might sell sometimes for, you know, $1,000 today, but when you're a new company, with a new band in a very new area, it's important that it's, you know, it's a kind of product, you can take your wallet out and say, Oh, I'll take a risk on that.

Yeah, definitely. And there's a lot of people that are willing to do that, because maybe they're way less that are willing to take the risk and like $1,000, robot vacuum cleaner versus a $200. One. Okay, and then where

we got, we got the robot on the market, the, the Roomba, we still doing the robots for the military. And that actually, we got them both on the market in 2002. The robots in the military got moved up because of the war efforts and Operation Enduring Freedom in Afghanistan, and then soon after evac, they probably would have stayed as military research for a while. But there was this Driving driving need for bomb disposal robots, and we were like, they're with the capability that they needed. And at the same time, we put the limbo on the market and you know, starting to get some consumer success. And so these companies never would have had these two disparate areas if one had happened first, right, because all the resources would have been thrown at that one, just because they were successful together. And you know, it really helped us because I can tell you a lot of the investors didn't believe in the consumer robot market until we had it on the market. And it was successful.

Gotcha. So that the investors were or was it like you were funding the efforts with the military robots. And that allowed you to bootstrap the concern?

We were we were paying for infrastructure with the military contracts, because we have large contracts as research contracts as well as having built up, you know, so the infrastructure was being paid for, like, you know, the offices and that we had people on staff to a very capable roboticist because we had the ability to pay for them. But we didn't really raise investment capital until eight years in in 1998. That's when we started. And by 2005, when we took it public, it you know, we'd raised a total of 38 million, which is peanuts today, that's like seeing these, you know, a OB today.

Yeah. Okay. And then you raise the capital, and then that lets you scale so you could grow the consumer market for the room, or the consumer side of the business for the room.

At the same time, though, you know, after we went public, the military side was going strong too. But eventually, after I left in 2008, they did spend the military stuff out into a company called andeavor, which was purchased by FLIR for like over $300 million. And then, that was purchased actually was all purchased by Teledyne. So the technology still exists, the robots don't exist, a lot of the same people are working on them, but it's no longer a robot.

Gotcha. What is your so between 1998 and 2008? Or so when you left? iRobot? What were you primarily involved

in? Everything I was, you know, going on time, I was president chairman, and, you know, everything from sales to media to taking it public and investment. Acquisitions as you know, everything.

Mm hmm. And then how did you decide to go to sci fi work? So that was the next step? Right.

Yeah. Well, I was looking at what next? And, you know, being chairman of a company is kind of the best job actually, what it really is, because it's a gotcha what I was at Iowa, but because you have, you know, you get to do a lot of interesting things. But you're not, you know, you don't have all that operational responsibility. But it's not the best job when you're 40 years old, which is what I was. So I think that I really, when I looked at what I really liked, I liked it when the Roomba and pack bought the military robots, were first being developed the innovation side when you're getting something the likes of which people haven't seen before, after. And so that's why I want to do it again. And I saw that the drone space was the place to do it. And

how did you see it was the space to do it? And what made

it was an untapped market. You know, when I look at drones, it's like, a magic technology that solves a lot of the problem. You know, I think it will guild guild fat who was adopted now at Toyota research, I think he he's the one who first said, The problem with ground robots is to ground all this stuff on the ground, right? Especially when you take it out outdoors, and even indoors with the Roomba. You know, there's a lot of pollen you have to face and a lot of recognition, you know, a lot of knowing what it is and getting around it. And negotiation was a drone, you could just fly right over it. Right? And you know, free space is actually easy.

Yeah. Okay, so you saw that there was potential here because it solves a lot of existing robotics challenges. So you could probably do new and interesting things. But then how sci fi works? And actually, would you kind of speak to the premise of the company?

Well, the Sci Fi works. The idea we had originally was at the time, a little less so now, but at the time, the battery technology wasn't there. So the drones would only run a short amount of time and not enough any real types of military missions or law enforcement missions. So or even, you know, long term communication missions. So we had some technology to make a tethered drone. And we were, you know, the first ones out there with a drone that you could stay stays up in the air for, you know, weeks at a time. So you put it up on a tether, and you put it up to you know, put it up to 300 feet in the air and it just flies all the time. And then you have that station where you can get eyes on something or you can do communications

never I never envisioned that the only thing that company would do, but I wanted some differentiated technology to go in because there were some drugs on the market already.

So the idea was, you'd have a cable that goes to the drone, and that cable would send power. And you could also probably do things like maybe some of the control, like the brains of it, maybe I don't know, if you want that on

putting the veins oil on the robot, so you send it up, and it just stays the or, you know, autonomously. And you know, what, that's a little bit turnkey today. But at the time, we were doing it, it wasn't so much. We are, uh, you know, that wasn't the the only tightness of company it was a starting point. And so we want some good military, you know, multimillion dollar contracts about that. We were working on getting it into an acquisition program. And I can tell you, there are many companies building tele drones today. So it was sci fi wasn't successful, but the concept was successful. And it's in pretty hot demand for the familiarity application applications because of the capabilities to do both surveillance. You know, I getting eyes on 24/7. And also, communications videos work a lot better when you put them up high. The effects of terrible on the transmissions,

yeah, but it seems to me like a really good way to bootstrap infrastructure. Like if you need to set up quickly, it's like, oh, we have a security camera that's deployed there through a drone, it's kind of thing or boost your radio up and put it real high. Like that's very nice. Very clever. Okay. And so you said, so a few things were interesting there that you said. So first, you had preparing it for acquisition? Does it mean you kind of created the company with the intent of it being bought?

Oh, no, I am sorry, I meant in that context, I meant military acquisitions, making sure that you're in the palm cycle in the military budgeting cycle. So that money exists when you're ready with the technology and prototype. Yeah, and at the same time, getting to a point where we got it deployed, we did get it deployed with a Rapid Equipping Force and will then be used in Syria and Iraq, very successfully by a special forces.

Gotcha. I mean, that feels really good to positively impact our troops and it both iRobot and and sci fi works. It's like you were contributing to some of the difficult situations,

it doesn't matter. Some sci fi, someone came up for Special Forces. Perfect came up at a conference and said, Hey, I heard that thing is out there saving saving lives. So you couldn't tell us much about the missions, of course, but actually, they did tell us I was having great effect. And the same thing that I wrote about once. You know, once I was speaking at the War College, and I was between two, you know, three star general, my vide and a four star general on my left, and I assumed everyone would want to go speak to them after the conference, but a lot of the soldiers came up to me and one guy, remember, he shook my hand and said, I want to give you my coin because this pack bought saved 11 Guys on one mission, and you know, that was really rewarding can be the Pentagon actually did a study and the pack blood should have been credited at that time, even more today. But at that time, they've been credited with saving the lives of hundreds of soldiers and 1000s of civilians.

Wow, that's absolutely nuts. It feels like it's kind of like, it's amazing. And it's probably hard to wrap your head around, like the abstract notion of saving lives for this kind of thing. But I it's just wonderful. Like, what a crazy thing.

And then so you said that sci fi works wasn't successful, in your opinion, or this kind of thing. Oh, we I,

I live in the we had a difference of opinion with the investors that I bought on in the direction. And at that time, I decided, you know, the person who first deployed the PAC bots over in Afghanistan, he asked me to come in and join, he became the acquisition executive, that's the person who runs the, you know, the, you know, multi billion dollar acquisition flows on all of them for the OMA and he wanted me to come on and be his special robot advisor. So I decided to take that opportunity. Because I didn't agree with the direction of cycline.

Yeah, that must be frustrating. I've heard many, many people that I've talked to that I started robotics companies where the investors wanted to go one way and the founders want to go what another way? It's a very frustrating spot. Right? Right. Uh huh.

But at some point, you got to say, you know, they have that opinion. They're gonna pick on that The people and if those people make it happen, the company's gonna be successful. If they don't, it won't. And I can tell you this one was really well set up to be successful because there's a really hot demand for the technology.

Yeah, and then so but it just wasn't where you were personally interested in taking it. And then this other opportunity popped up to advise on government purchases of robots, right?

The government research and development of robots AI and machine machine learning.

That's super cool. So that was, so I got to

work in the Pentagon for two years. Which was great. I mean, I, you know, I stayed in Boston, where I reside, but we went down, though, every few weeks to spend days down there. And, you know, my office was the Pentagon?

That's so crazy. Yeah. How was that? Like? I don't I don't know much about it, or what would be involved? Like, what? I don't know, how was it?

Well, after selling, you know, over a billion dollars worth of stuff to military, right, it was interesting to be on the other side of the table, and see some of the, you know, constraints they're under, right? I mean, there's a lot of hard working, patriotic, very, very good people, and you say, why can't they just do this? And, you know, sometimes there's a lot of reasons that they just can't do that, you know, mostly, you know, it's against the law.

So, you know, give me a specific example.

Well, you know, even if they want the technology, it has to go out for competitive bid, and you have to have the evaluation criteria, and you have to, you know, go through the process, because someone will complain like that, Congressman, and then there'll be a bit dispute, and it's better, especially when you're dealing with billions of dollars to get the program right from the beginning. So you don't have to, you know, lose it, you know, later on, because the, the, the soldiers need this equipment, right. So, sometimes, you know, even if it's a really, really good product, and you know, it's going to be the winner, you really have to go through the, you know, the process.

Gotcha. either. Yeah. So that's interesting, it's interesting to do that, that's very different than like how startups might operate, or this kind of thing, because you don't have the implications. Like the voters and all of this, if it goes badly, you just like, do something and fail and try and whatever. It's interesting to be on the other perspective. Did you? I don't know, like, any. Like, I was very impressed. So I talked to a previous episode, Tim Chang, from DARPA, about the sub D challenge. And I was really impressed with that. The like, it seems like a very pragmatic approach in robotics, to these hard challenges, and I'm trying to start these communities around around this area, I should

say, DARPA does have a little more leeway, because of the way it's set up to be able to work with universities, and, you know, get those great new ideas into the military. But when you get to a, you know, a knowledge acquisition program, it has to be, you know, go through a process.

Yeah, so were you kind of responsible for advising these large purchases, or I wasn't funding a process or

I was brought in to give an honest opinion on the state of the robot technology. And I didn't take a role in any particular acquisition program. But I think people kind of thought of me as a trusted adviser, and I can't talk too much about it. But I think I really did have a great impact and getting some of the, the drone and the ground vehicle programs in the right direction.

And how did how did that end? So was it just the end of the term? You have a two year term? Or how does it

Yes, I I actually went until my first appointment and out and they would have had me come back, but especially with COVID. Like, I couldn't even fly down to the Pentagon anymore. And, you know, I, I was at the point where I felt that I'm doing something else to push the robot field further would be what I most want to do. I see.

Okay, and so now this comes to Tertill, right. That's the next jump. Okay, now, tell me a bit about Tertill. Just the set up in the premise and everything.

Okay, so Tertill is a company that builds a weeding robot. It was invented by the gentleman that I mentioned, Joe Jones, who was invented the Roomba when he was I know but and he, him and Lloyd McKee, again, he they were out of they came out of harvest automation, which was an agricultural company. And once they had finished with that, they decided that they wanted to be back in the consumer game. And, you know, Joe, we're thinking about all the different applications that have robot potential and weeding, you know, is really at the top of a lot of people's lists those jobs that are just tedious and hard, and, you know, you're in the hot sun. And it's a job that, you know, you do once in a while, you know, find somebody who likes it, right? Just like you're fine. Believe it or not, you find people that do like vacuuming. There's always people like that. But, you know, it's, you know, a lot of people like it, but they like it on their schedule, like not when it has to be done. So Joe, thinking about the problem, he looked for a robotic solution, because, you know, we believe a Tertill that an AI robot that you don't have to do things exactly like a human does. In fact, to be more powerful, you don't really want to do things exactly like a human, you want to do them. You want to set up the past, so a robot can succeed. So an example is, you know, maybe Roomba can't do everything that an upright vacuum does. But it should does well going on to the beds. Something that an upright vacuum doesn't do well, doing under the couches and the robotic applications, it doesn't matter how long it's on the you know, covering the whole space, and now with the latest technology generations, ensuring that the space is covered. So you know, when you think about it in robot terms, instead of in human terms. So one way to do the weeding to get back to total, one way to do the weeding would be you know, you have a fully dexterous manipulator and you, you know, you use your AI vision system with lots of machine intelligence, and you say that's a weed, and then you come down and you, you know, you have your dexterity to go down and grab the wheat and pull it up. And then you have to grab it and pull up the, you know, the roots. Or you use events, lasers or something like pesticides. And in thinking about that, that's a long way, right? Because that plant has already sunk nutrients out of the soil, you have a disposal problem with the plant. And you know, quite frankly, powerful lasers are expensive and pesticides.

we always consider lasers, because you know, having a robot with lasers, it's just inherently cool. But may not be the most practical approach. So what was built instead was a robot that prevents weeds from ever growing. So with a scrubbing, scrubbing wheels, it keeps seeds from germinating. And with a spinning weedwacker. If one does pop its head up, off with it. And original Loomba navigation strategy, it negotiates through the garden and covers the entire area using local intelligence. But it's got some inventions on it. It's got capacitive senses that sense the plants, and it's got a very innovative pump sensor that uses the accelerometers.

So it actually, would you just for our listeners who are only listening, would you describe it? Describe Tertill a bit.

This is a small, round green robot, it's about eight and a quarter inches wide and about maybe four inches tall. And it's it's got wheels that have cleats on them. Four wheels that are a camford with cleats on them that makes a very,

you know, just at an angle that 45 degrees. Yeah, tilted so that they kind of like scrape the Earth. Yes. And in

fact, we build a not very good die system because we want to be scraping the earth when we turn rather than lightly travel. Like many robot applications.

That's so funny. So you turn all the wheels

I think they skid steer it's really the it's really the combination effects of scrubbing and then the weed whacker that makes it do such a good job.

So then in the middle of the robot, so you have the wheels and then in the inside of the wheels in the center of the robot there's a little tiny spinning thing that has like a wire like a weed whacker.

Yes. It's not just like a weed whacker. It is. Yeah. This is the only thing Part of the robot that that you have to change, so maybe three times a year, you have to pull it out, within a, just do it like that and you push it a new one because it does wear out when it hits locks and other things in the garden. It does have a little USB port here that whoops, does have a little USB port here that if you take this off, you can see a little mini den, I think it is where you can plug it in and charger. But the great thing about this robot is you never have to, because it's got a solar cell on the top. So you're putting your garden at the beginning of the growing season, and you leave it there and every day when the sun comes out, it charges up when it gets down to you know, when it was when it goes out and weeds when it gets down to something like 70% battery, and these all changeable numbers, but 70% battery, it stops weeding, and it charges again, and then when it gets to a certain number, it'll come back out and do the do the weeding and then when it gets down. And that way, we really use the batteries really lightly. So it lasts for years. You know you're using them so, so nicely. And it makes it run throughout the day. So if it's, you know, if it's cloudy in the in the morning light, it'll, you know, keep running in the afternoon, it also has some very intelligent algorithms. So I'll go sit at the sunlight, rather than under a tomato bush. So it's really a very intelligent little robot that lives in your garden runs on sunshine, and does the weeding so you don't have to

Gotcha. So it drives you basically set it in your garden, and just leave it and then what it's going to do is charge period charge with the sun and go knock out some weeds when it drives around for a small period of time each day,

individually throughout the day, so it doesn't just come out once it comes out a few times every day and does the job.

Gotcha. And so how much ground can this little robot cover?

About 200 square feet of a garden. So that's about you know, an average size in ground garden. Some people have smaller ones these days with a you know a vase bed but you know, if you're putting in a garden you you know you have to so it covers a reasonable sized garden.

Gotcha. And it detects the weeds to kill because they're small enough to drive over.

Well, it doesn't really detect the weeds or kill it prevents weeds from ever growing. What we have to do is protect the plants. Now, you know, as I mentioned that capacitive sensors are very good and the accelerometer based bump sensor is very good. But if something's less than four inches, it will consider it a weed. So we do what all good roboticists do to make robots work and that's cheap. The way we cheat is if you put a seed in, yeah, well, I am what what I'm holding is a thin wire that's molded into a curved shape that you can put a round put around a seed that you planted a small seedling that you plant by just pushing it into the ground. If you plant a row of carrots, for example, we have a row God and you just push this these along the row so the robot can't get to your carrot seedlings. And once the plants four inches tall you you can take these out or leave them in and it's really your choice to take them out. You could just be used them.

Gotcha. Okay, so when it's driving around, does that mean that it's what weed whacker is always going? So like Does it smell plants? Or does it?

It can sense the plants, we've discovered that having a duty cycle does actually help the performance so we can send so we can just have a duty cycle just Yeah. And get them all. Gotcha. It doesn't take that much. Damage power, especially when it's not hitting anything.

We don't just keep making it's got a duty cycle.

What do you mean exactly by the duty cycle? It's on for a little bit.

It's gone for a little bit. It goes off? Yeah.

Oh, is it? Okay, so it's base. It's like almost like it's full power than offer a little bit than full power than offer a little bit while it's driving. So it's

yeah, you need to have a weed whacker. You need to have them spinning like a very high speed weed whacker. Right.

Gotcha. And then, so what what is what exactly is it sensing in the environment? So you mentioned that it can move into the sun, if it's in the shade.

That with its solar power system light with a solar panel, so

it's going to drive on till it reaches some sun kind of thing, or how does it work?

It? Yeah, it just doesn't stop. If it's too shady, it keeps going for a while and it you know, it looks at the average reading over time and goes back to, you know, web it, it looks a bit if you know, wants to know, make sure there is some sun before it starts looking for sun. So yeah, if you have an average eating and, you know, none of these things, you know, these things are tuneable. So we've done the best job we can, you know, tuning them that we that, you know, that we could do, but they are things that we can vary if we need to at any time.

Yeah. Okay. So I am imagining that this takes a lot of lessons from the Roomba. So like, one thing that was very obvious to me as a like Roomba tie back, is that you have the one button. So like with the Roomba, I have heard that it's like you just have the one button to start it. And this Yeah, it has the one button as well push it. And that just goes and you don't have to worry about

it didn't always have one button, right? That was a lesson learned. We had three buttons, and they were labeled S, M and L. So small, medium and large. And we get questions all the time. What's a small room? What's a medium? What's a large room, so then we started putting numbers out there, then people are really confused because they don't know the area. And so it was a little it'd be much better to have one button. You know, now they say clean on it. Right? So that's what you want to do you want to clean you press one button. And the on off switch the first ones we had on the room, but were labeled zero and one. So one is one is on and zero is off. But you know, people would say well, I put it on. Oh, isn't that on? And it was not obvious to most people. 40 year old

Well, it's kind of funny that we would do that. Like, yes.

I mean as a roboticist. Like, it's like, oh, yeah, I got it. Like, yeah, I know what you're going for. It's like all you but it's not for all of the other anyone who's a general consumer, I see what they mean with zero seems like Oh, fraud. Like, it's totally it's off to so.

Yeah. Yeah, that's what the confusion was like, why? Well, what about off? Oh, so we designed it with one button and the one button. You know, it's, it's something that people can understand, like, you put it in the garden, you press a button, and it stays there. And it's on and it's running in the garden for the rest of the, for the growing season.

And then so, so the operation of this as you just put it there, you push a button, and then you walk away until winter, I suppose

that would be ideal. That would be ideal. But in reality, because we live in reality. Mine and even though I made my space pretty. I mean, I am a little bit of a lazy gardener. So my place isn't really nice. And you know, beautifully make do anything like that,

if I had to free it a few times during the year, like it would get stuck, and I would give it a kick. But that's a few times as opposed to every day, you never have to charge it. You do have to replace the weed whackers every, you know, every few weeks, but apart from that, you know you're good to go for the season. So it really is, I'd say one of the most if not the most autonomous robot out there because it's solar powered. You don't have to charge it. And it just stays there and does the job for you. Like a little creature living in your garden? Like a Tertill? Yeah, like a kettle?

Huh? Yeah, I see. And so you mentioned that the control style or the control scheme is a bit like the original Roomba. Does that mean it's gonna just drive around randomly? And then eventually, like in the limit, it will have covered the whole garden

space Yes, although it's never been manned. And the moon does whenever random a us you know, local intelligence based on the the, you know, the local sense of meetings, to negotiate along walls about obstacles and to try and cover you know, an area and then move to the next area. So neither you know, the original room has went random and this one's not random. We don't have a full positioning system yet. And you do have to put a boundary around the area where it's working like there was a luxury in the house that the walls originally that pretty much we had a boundary because of the walls. Turns out most vegetable gardeners have a boundary or they don't mind putting one in. You can take a piece of wood you can take the VO Gods the pieces of metal that it comes with you can buy at stripping a metal one from From Home Depot or Amazon, you know, there's a lot of easy solutions for it. But it is one of the reasons why we don't advertise it for landscape beds, because it's a little tricky to explain where it can and it can't work. And, you know, you do have to have a boundary of bounded and some landscape bed, that's fine, I've got one in a landscape bed in the front of my house. And it's wonderful, I don't have to weed it. It's amazing. What I didn't mind having, you know, when you have, like, you know, you know, some beautiful shrubs, I have some bamboo in a, you know, with a bamboo belly around that most people can't stop bamboo from growing, but my bamboo isn't going so everything else was going in that bed. So I put, I put a barrier around the where the bamboo Spedding farm and I'm giving it a chance to go in. I also put one on my gravel last year, you know, got pea gravel. And we can't advertise for that either. Because at the end of the season, I'd look down, there'd be some little portly plants growing that had established a boot structure. So that type of weeding we do to venting, which doesn't quite work. But where I didn't have the Tertill weeds were like, literally up to my shoulder, like up to my shoulder. Like it's crazy. What was how fast the things could grow in a growing season if you never take care of it. So for me, you know, it works great. If you don't mind a few little leaves in your in your gravel. But I you know, advertising it for that would be would be hard. But we have a lot of users who use them on on landscape beds, because they figured it out for themselves that I meet the criterion. And so we can have it running. And it's not a problem. You know, you got to have your plan set of space, at least about a foot apart. You have to have not too much. Exactly, yeah, yeah. So the robot can go right between them. You want to have not too, not too much undulations a barrier around it. And it doesn't work on really chunky mulch, like you know, a lot of landscape beds are mowed. And that prevents weeds. I put motion fresh and even last year, about a month later, I got weeds all over. So it's really chunky motion. It won't work. But I have it working on both beds that have final motion. It works fine.

Gotcha. So what have been some of the biggest challenges with this robot it? I'm imagining weatherproofing it it's interesting to have the small set of sensors. I don't know.

Well, I actually jumped into this. After they kick started it, and I bought one on the Kickstarter. So I can do this. Because my vote because you know, Joe started the company and I knew him and some of the other guys and of course, I'm gonna support the Kickstarter. I, they delivered it. And I you know, I put it in the garden. And I think, you know, it worked for a little bit then it stopped working. And I didn't care because it was a Kickstarter, right. And it's a robot and I buy lots of robots on Kickstarter. And quite frankly, I like that. Oh, yeah. Yeah, anyone could tell me a robot on Kickstarter. Because I want robots to exist. I want more robots to exist. Yeah. So everyone should always support the small robot companies just you know, get the robots try them out like that. So many people said that to me about Roomba. Like I got one of the early ones and look at the generation they have. It's like, no, no, that's 10 years in. You need to get to these companies when the first kick started when they're, you know, just have it on the market for a few years. But I was anyway, I got the robot on the Kickstarter, but it stopped working. And I didn't really care. So but they actually came over, took a look at it and replaced it. And I then I put it in my vegetable gardens. And I was actually amazed that it worked. It was like night and day, right? It was like instead of going to my garden, and it was a weedy mess, because I you know I'm not one of those people who does it every week, you know, on time. Instead of going there with my daughter and you know, I get distracted because I don't like the weeds in it, you know, instead we can go and garden together right? Instead of you know, instead of it being a chore it's like a more fun experience about growing your own food. So I really loved it and they I was looking around of what to do after the you know, after the military appointment and you know, I called them up because I you know my mind I was thinking about outdoor robotics because you know, I'm not going to go into us right because we will visit inside I was thinking, Hey, I see the robot lawnmowers taking off, you know, I've got a Tertill meeting my goal. And that might be another space that you don't they've maybe thought about that and not pursuing and I could take that and learn from them. And instead, they invited me to come on and be CEO.

Tertill's Website

Helen's LinkedIn

Tertill's on LinkedIn

Solar Powered Robotic Weeding

Cheating, Like a Good Roboticist

Saving 11 Soldiers' Lives

High Powered Lasers

Helping the World

Bullish on Robotics

In this episode, Audrow Nash speaks to Dayo McIntosh, who is the founder of Yateou; Yateou is an early-stage startup that makes wellness products and has a customer facing robot, ADE, that personalizes customer orders. Dayo speaks about what motivated her to start Yateou, how she uses the robot arm, ADE, and about her plans for Yateou's future.

Introducing Dayo, Yateou, and the ADE robot

How does ADE work

Putting ADE in a truck

Keeping it simple

Having ADE stolen!

Customer experiences so far

ADE and the novelty effect

Deciding to use a Kuka arm

Dayo’s background

Getting into beauty products

Finding a place for ADE

Focusing on the customer experience

The value of self-care products

Yateou team

Yateou’s direction

First hires

Yateou’s funding so far

Lessons learned

Contact info

This is a conversation with Dayo McIntosh, who is the CEO and founder of Yateou. Yateou is an early stage startup that makes wellness products and has a customer facing robot ADE, that personalizes customer orders. One thing that I found cool about Yateou is that they put it on a truck that's kind of like a big food truck that has a big window so that customers can see it working to make a customized order. It's an interesting way to draw people in so that Yateou can start a conversation with them about wellness, which is where Diana's interest in the idea started. I enjoyed talking to Dayo to hear about her experience with Yateou. Her background is in robotics. So it's interesting to hear about how she came to using her robot arm ADE, as well as hearing about the challenges she's encountered to get to where she is, such as having her truck with the robot stolen. And it's also interesting to hear about the clever things that she's doing to manage the complexity in her growing new business. I'm Audrow Nash. This is the sense think act podcast. Thank you to our founding sponsor open robotics. If you enjoy this interview, please consider subscribing. And now here's my interview with Dayo. Hi, Dayo, would you introduce yourself?

Hi, Audrow My name is Dayo G. McIntosh, and I'm the founder and CEO of Yeto.

Yes, yeah, so is this sustainable clean beauty and wellness brand. We create products that are all natural vegan, cruelty free and ethically sourced. And the unique thing about what we do is that we've incorporated robotics to bring that technology front and center to the beauty and wellness industries.

So right now we have our essential oil mixologist robot called ADE. ADE is the first consumer facing robot for the beauty and wellness industries. And currently we have ADE stationed in a moving truck, making it mobile so we take Addie to different events, markets and things like that. And people can interact and interface directly with ADE to create customized and personalized personal care and wellness products. Gotcha. What,

Addie is a Kuka robot. It is about two feet tall. And it's a manipulator. It's an arm. It's a robotic arm. And again, it sits in a truck with a transparent glass that people can see. See and watch ADE in action.

Gotcha. So you said Abby, Eddie's the next mixologist. What does that mean? What does that actually doing?

So similar to a bartender, right? We have ADE going to different dispensing vats to create products, again, tailored to each individual person's like personal care and wellness goals. And once. So for example, we have an interface and someone like you would go up to the interface, for example, a screen a tablet, and say, hey, I want to mix this ingredient with this ingredient. And I want it to smell like this. And that would kick it into gear. And that's how we create personalized products.

Super cool. So for this, is there any like? So if I say my skin is super dry, and I want it to smell like oranges, or something like this? Does it? Does Addie come up with a recipe or dive to say I want to this much I don't know, shea butter in it or this much whatever. How does that work? Is it recommended based on goals? Or is it like I know the ingredients.

So the full picture, the full rein skill of this is to have give consumers the options to customize and mix and match however they want. But rolling out we're going to start out with pre made orders that people can adjust. So it's more of an adjustment than a full blown customization feature.

Yeah. Oh, gotcha. And how does so for ADE to do its mixologist work? How does Addie actually go to the dispensers or other dispensers robotic to does that he push into something and it comes out like a soap container or how does it work?

So the dispensers are not robotic. So they are actually wine bottles that we've added dispenser or would you call it droppers to them sort of thing. And the idea behind that is that we wanted to eliminate as much or I wanted to eliminate as many possible holdups by limiting the number of automations or automated pieces we had in this operations. So ADE is the only robotic automated piece in this whole setup, and it goes to each bottle depending on what you select and pushes up on the dispensing nozzle, that's what I was looking for pushes up on the nozzle, and that releases a set amount of liquid ingredients for the product.

So it's kind of like one of those soap dispensers and restrooms where you you push the thing and a little bit comes out

exactly, it's exactly like that, or something you'd see at a bar, for example, I don't know if you've seen those. Those dispenser actually, I think they have them now at home, like you have those mobile bars that you can install in your kitchen. And they all have like nozzles that you attach to the bottom of, of Yeah, like a cocktail bottle or liquor bottle, and then you just push up on that, and then you get what you want.

Gotcha. And then how do you how do you control Addie? For this? So does it does that he know the position of all of the different ones. And you can say option one, option two option, it will go push option one, option two, option three.

That's exactly how it's set up. Yes. And currently it is not fully automated. And that's what I'm working towards. So for the time being, we use the Teach pendant, what we call a teach pendant. It's sort of like a controller. So think of like a game, a game controller, for example that you use to with your console. So instead of that, like we have this teach pendant, and I can call out the different positions that I want ADE to visit with the Teach pendant, and that keeps adding together.

So how does the Teach pendant what I'm imagining is like a big pen kind of thing with maybe like a bunch of QR tags or some some fiducial markers on top and you like hold it into a spot and there's a camera looking at it and you click it or something? And it says okay, here is the location I want? Or is that how it works?

No, think of a really industrial looking tablet. Oh, okay. So that's exactly what the Teach pendant is the controller. And so like, on this tablet, we have all the different programs, we have different positions. And if I want it to go to bottles one and seven, I enter that into the Teach pendant on to the Teach candidate and make that selection and click press or go play. And that again. gets it going.

Gotcha. So you select all of the things you want it to do on this. Exactly. It's like I want I want it to go to one and seven, whatever. And then it will grab the bottle, bring it over there. How do you get it to know the initial positions of the bottles where it has to push to get the stuff?

So if you if you have all the dispensers up? How do you? So if you add a new dispenser to the right, say a bit further away? How would you tell it that there's one over there that it could go push for option 10? Say

so right now everything is controlled, we have a limited amount of dispensers, we have 20 of them. So 20 positions that are set. So if we're introducing a new ingredient, we're taking something else out. So like it doesn't go beyond the set parameters that we have in place right now. And

warming up for the first ones. So for all 20 of those. How did you tell Addie all those positions.

Um, so that was programmed. We programmed it. And I worked with engineers to come up with a programming for this. And so that was hard coded. Yeah. And so adding those position one is, you know, to disagree. Yes. Away from Me. And yeah, so you can get to all those positions. Without knowing what exactly is there, right. It's just going to position.

Yep. And how does so if you want to do you want to make a custom mix for someone? Is it kind of like these different dispensers will top off the bottle. So you have like, maybe you can pick the framework fragrance and that's like one dispenser, and you can add some extra attribute like extra moisturizing or whatever it might be. And that makes it a little bit custom but is it largely made already, and you're just adding like a little topping to it.

Exactly. It's largely made already and what we're working on right now So you can make adjustments to the, to the recipe.

Gotcha. Very cool. Yeah. So and then this is all in a truck, and that truck has a big sheet of glass. So you can look and watch the robots working. Can you tell me a bit about that?

Yes. So when I set out the goal, and again, the vision still for this is to be in a brick and mortar store, to be sort of like the central point where people can come together. And it's going to be a safe place for self discovery for exploration, like finding clean, natural ingredients and products that work for your hair, skin and wellness in general. And so that was the goal. But living in the Bay Area, we all know that real estate here is really expensive. And so when that didn't seem feasible, I had to think outside the box, and that's where the idea of the trucking came about, I really liked the idea of having this experience, right, we're building this experience for people that will sort of engage them and get the conversation going about what they're using, in their, like, on their bodies, for example, and what their overall personal care and wellness like regimen. Resume is. And so that was the idea behind this. So you walk up to the truck and and immediately you get this experience, right, you're walking up and right in front of us is robots, you know, working to create a product tailored for you.

And so the way that it looks for the customer is they come up to you and your truck or whoever you're working for it. And then they talk to you for a little bit and then you poke on your teach pendant or whatever you call it a tablet. Yes. And then do you do you hand ADE a bottle? Does that ego pick one up? Or how does it work?

So right now we have it set have different trees, the different bottle sizes in the truck. So when we make the selection, it knows what bottle to go to based on what we've selected. So if we pick a small bottle, it will go to tray one. Medium bottle tree two, large bottle tray three. And so that's how it's currently set up. Yes. So we don't do any of the like handling of bottles or ingredients like everything is housed again in the truck. And ADE manages that whole setup. So the same?

How does that he pick up the bottle? Because that seems like a kind of tough task. So you have some sensors, like you have a camera or some sort of like physical sensors or something for telling where the bottles are?

Or how do you know. So that's something that we're putting into place. But right now, everything is hard coded. So we can't afford to, you know, move anything. So those trays are bolted into the ground so that regardless of where we go with the trust, like nothing moves, and so that position is always constant. So it always knows if they want a small bottle, I go to this position. Open up my gripper and there's going to be something there for me.

Yeah, but so getting a bottle on a tray. The tray is fixed, but like it has a bunch of bottles on it. Correct? How does it go and grab a bottle on a fixed tray? Is it just you have like a bunch of bottles sitting in kind of like, I don't know, separate sections for each bottle and it goes in it knows that this one is empty. Now it goes to the next one.

How does it sell a bottle? So everything that we've done right now has been like really rudimentary, right? eliminating any, any potential? Yes, any uncertainty or any

possibility of vacation for small, medium and large and you replace it every time or

Yes. So we have a long tray and I feed in the bottles or whoever feeds in the bottles. And then we have it on an incline. So whenever ADE grabs up, it slides down. Exactly, exactly. It's clever. Yeah.

I really liked the approach you're taking, which is like, keep the problem as simple as possible, because you can always grow and there's no need to like automate everything, especially while you're testing out an idea. Right. And so I really like it that it's so clever because you have three spots and the bottles just slide down into it.

Exactly. So one of my biggest fear when I started doing this, I did a lot of research about a lot of like consumer facing robotic applications, right. And I found that the number one problem that customers or people seem to speak about was that these different applications they were always down like they weren't. They weren't working. They weren't working often. And the problem with that is if you have so many are different stations that are all automated If you run the risk of like one or the other going down, and so you're constantly fixing a problem. And so when I set out to do this, I wanted to simplify it as much as possible, again to begin with and figure out the rest as we, as we grow as we as we move forward. And so like everything is super simple. What I'm building right now, yeah, yeah, it's it's been good so far.

And how so Okay, now ADE has picked up a bottle from one of a set of spots. So you say a small medium or a large bottle, or whatever the sizes are, she grabs one, and then pushes and adds a few different things from the dispensers. I assume the bottles have no tops. Yes, correct. So you can squirt right into it. And then does ADE like, is there like a little door or something where ADE puts the bottle for you to give it to the customer? Or how does that like the whole process now?

Okay, so let's say you walk up to the truck, you're like, Hey, do I want this product. And I wanted this way, I input that selection onto the Teach pendant that keeps adding together, it moves over to the trays with the with the product bottles grabs a bottle, small, medium, or large, takes that and goes to the dispensers on the wall. So if you selected product number one and seven, for example, it would go to product one, or dispenser one dispenser seven. And once that's done dispensing, it would deliver the bottle to two cutouts that we have in the truck. So in the truck we have we cut out two doors, think of it as a trap door, two doors with a revolving revolving tray. So as you would place the bottle on the tray on the inside of the truck, we would revolve that tray and it would appear on the outside of the truck for the customer to grab.

Cool. Okay. So, with this, how long have you been doing this? So how long have you had the truck setup with?

Ooh, two years. Two years? A little over two years? Yeah. So we started doing our beta and showcasing it in q4 of 2019. And then, as we all know, COVID hit right, the next year. And so all of that shutdown. And actually it was stolen. And so that set us back. And so we've been in the process of like rebuilding. So that, you know, as things have opened back up, we can get back out there. I'm showcasing and doing what we do.

So you don't currently have an Audi replacement ADE.

Yeah, it's a funny story. So now I currently have two ADDYs. So the original ADE was still the truck was stolen with Abby. Yeah, they It's wild. They ended up destroying the truck. Taking the computer, the brains behind it. So it is a manipulator, right. It's just the arm, the hardware, and then you have the controller the brains behind it. So they took that I'm guessing they took it for the wires. So that was missing. And so I have to start over purchasing you ADE. And so now I just have the manipulator the arm for the first one.

That's crazy. I can't believe just how did they steal your truck? With this? Did it like at night? You just found your truck missing or Yeah, kind of thing? Are you in the Oakland area? Or were

you? No, I'm actually I would consider this the suburbs. I live in New work close to Fremont and just across the bridge from like Menlo Park and Palo Alto. So yeah, that that was devastating. Yeah, like I said, it was right before the lockdown. So no one was working. Everything was shut down. And so we just started rebuilding at the end of last year.

Gotcha. I just had a friend who had his car stolen, and they just found it recently been so crazy that

And my robot truck. Yeah, like a whole truck. I was like, how do you steal a whole truck? But yeah, that happened. So it's part of the year two story.

Gosh, now, so two years a lot. Some have some difficulties with like the truck being stolen and COVID. But how have customers been engaging with ADE and how has the experience been and the times that you have been in front of people?

So um, so we debuted ADE unofficially at Afro tech. So the time which is November berth 2019, it was just a really great opportunity to, like, showcase Addie, and have a crowd of people where we could get feedback and just kind of gauge people's reactions. And so that was a really awesome experience, we got a lot of feedback, we got a lot of interest. And we were really excited to launch after that, based on the interest and the excitement that we found, or that we received, being at that event. And yeah, so it just, again, all of that, the lock downs, the theft, all of that set us back, but we're gearing up again, to go back out there. And we know that we're going to have the same amount of like excitement and interest that we had initially. So that's what motivates us to keep going. And that's what we're looking forward to.

And the any like funny stories about people being just like, thrilled or anything when the robot gives them their custom product or anything.

So when we launched, we weren't actually or when we did our beta back then we weren't actually in operation. So everything is just like, online right now. You can buy premade products online, but we haven't actually ever started selling via adding those evolving demos. And again, yeah, we were we were all of our plans were, you know, shifted by but in disarray, by everything that has happened. Yeah. When

So we were really excited to again, have this experience of actually selling with Addie and gauging customer experience, user experience. And, again, getting that feedback. I think that's really important. As we iterate and as we grow. Right. So

what So you mentioned it a little bit, but the hope with Addie, is it start? So basically, people come for the robot, and then it engages them in meaningful conversation about their self care effectively. Yes, that's yeah, the draw of the robot and then maybe build a community around this. Yeah.

Well, so at ease, definitely a novelty, right. We haven't built into a truck. And I don't think right now is with the right love until like, it makes me super excited. And and robotics is the future of so who knows what will happen. But it is a novel team. And the idea is that the efficacy of our products will bring people back to say that it serves two purposes. First, it's experiential retail. So again, you're having this experience, people have this fun, it's you're having conversations, you're having community, right around this really fun, this novel, really interactive, unique experience, right. And then the other hand is operational efficiency. So we envision a time where the second it will be put to use in the back fulfilling orders and manufacturing. And so that's that's the goal, the single robotic arm can turn out based on our calculations, let's say about 500 products in a 24 hour period. And even more than that, so that definitely kicks into gear, what we're doing, and we're able to put out huge volumes, as control all of them.

Doing it manually. How long would that take? You to do?

Yeah. I mean, 500 Do you have someone like making that? I would say maybe, like less than half of that. Possibly. But then you don't have people working? 24/7 Right. Like,

definitely. We manual and repetitive and like yeah, not

rather, exactly. would rather people would rather the humans use that time to engage and interface with customers and have meaningful conversations versus doing this manual? Taxing indexing work?

For sure. Oh, what a cool thing. Yeah. Let's see. So how, I guess how did you pick a Kuka arm? Like those are? Like, I believe pretty expensive arms? They're like really nice arms? Yeah. I think like, I mean, you don't have to give me the exact number but I think like $40,000 for like one of those arms. Something.

That's That's exactly right. They are quite expensive. That is a very good question. I have to this day I I don't know I just have this fascination with Kuka specifically have this fixation on Kuka. And it again, it's unexplainable. I don't know why. But I think the few times I've been around robots, they've been cool. Cuz so that may explain that. But as far as the cost, you're absolutely right. They're all super expensive. And for someone who's like bootstrapping, And so big expense all at once. It's a big expense. And so when I set out working with limited funds, I send an SOS and this is what I call it an SOS different robotics groups that I was a part of asking if someone if anyone had any, like, used robots, you know, that they could sell sell to me for a certain amount of money. And someone happened to respond. And at the time, she was the CEO for Cafe X, the coffee robot in the Bay Area. Cool. And she just responded and say, Hey, we have this, this arm in the back. It's been there for years. We don't even know if it works. But you're more than welcome to have it. Yeah. And so that's how we got started. So I got that for like a fraction, like maybe 10% of the cost of what so what else? Yeah, yeah. So that's, that's my wild, wild acquisition story. The very first

2 posts with the consumer tag