In this episode of the Sense Think Act Podcast, we are joined by Andreas Bihlmaier, Head of Robotics Software Architecture at Intrinsic, for an in-depth discussion on the future of robotics. We explore a range of topics, including the development of adaptive robotics software, the potential of cloud robotics, the concept of robot app stores, and the challenges of testing robots in real-world production environments. Throughout the conversation, Andreas shares his expert insights and sheds light on the innovative technologies that are shaping the next generation of robotics. Tune in to gain valuable knowledge from one of the industry's foremost thought leaders.

Start

Introducing Andreas

What does a Chief Software Architect do?

Coming to ABB

Planning for new technology to advance

Serverless Robots?

Robotics and other technologies

Growth of the Robotics Community

Robot App store?

Tesla bot?

Human in the loop for startups?

Reinforcement Learning in Robotics

Testing Robots and how it’s different than other software

Learning about testing

Advice for those early in their career

Getting in touch

Andreas Bihlmaier's LinkedIn

Andreas Bihlmaier's new book: Robotics for Programmers

Intrinsic's website

ABB's website (Andreas' previous employer)

Production-ready Robotics: How to Move Fast and Not Destroy Your Robot

In this episode, Audrow Nash speaks to Tobias Holmes, Quality Assurance Manager at Blue River Technologies. Blue River uses computer vision and robotics in agriculture and was acquired by John Deere in 2017. Tobias speaks about herbicide resistance, spraying weeds, quality assurance and testing on hardware, and on encouraging kids to learn robotics.

Tobias introduces himself

About Blue River

Agriculture 101

How non-row crops are handled

Row crops and herbicide resistance

How weeds work

Strategy for row crops

Genetically modified crop

Blue River and John Deere

What Blue River does best

John Deere acquisition

Timeline for Blue River

What it means to Manage Quality at Blue River

Quality assurance in hardware

Machine learning in weed spraying

Testing systems in the fields

Thoughts on robotics and agriculture

Using AI in mechanical design

Encouraging kids to program + learn robotics

Advice for students

Links

This is a conversation with Tobias Holmes. Tobias is a Quality Assurance Manager at Blue River technology, which uses computer vision and robotics and agriculture. Blue River was acquired in 2017. By John Deere, we begin this interview with an introduction to agriculture and the problem that Blue River is working to solve. This includes talking about row and non row crops, and herbicide resistance, both of which I knew very little about before this conversation. After establishing the problem, we talked about how blue river systems work, quality assurance and testing and hardware, machine learning and weed spraying the acquisition of Blue River by John Deere, the future of robotics, in agriculture, and on education. This is the Sense Think Act Podcast. Thank you to our founding sponsor, Open Robotics. And now here's my conversation with Tobias Holmes. Would you introduce yourself, Tobias?

Hi, I'm Tobias Holmes. I am currently the manager of quality at Blue River technology here and in Sunnyvale, California, a subsidiary of John Deere Corporation. And, and yeah, I've been working in robotics for about seven years now.

Would you tell me a bit about Blue River?

So Blue River, the company started out what what our co founders, Jorge Harada, and Lee reading. A couple guys from Stanford, decided that they wanted to bring high tech, you know, high tech tech high technology into the agricultural space. So years ago, probably about 10 years ago, they started out going carried harvesting. And you take what's out there harvesting, Carry, carry weeding, sorry, carrier weeding, and in 30 days, yeah, so basically being able to look at a carrots and be able to clean up the weeds that are around the character that the character actually gives you more read more of the harvest, you get more of a yield from the harvest. And and that was something that that was interesting, there was a problem that was happening in the world, they were in the tackling that using various robotics technologies, and talk to a girl who was like, Hey, you guys have a nice technology here. But what I really needed to be utilized is in leading of row crops, and so Blue River transition from

What are row crops? I don't know that much about agriculture. And I don't know about how carrots maybe are not row crops. What's a row crop?

So row crops are when you typically think about a farm and when you're on a farm where a grower grows, they grow their crops into these like, you know, very uniform rows across their entire field. And that this allows them to get the maximum amount of yield from a particular crop you're trying to grow, it maximizes the space that you have, in order to to get the most out of it.

So you just plant them, like optimally spaced in lines. And why why are carrots not row crops? I would just assume everything is planted in a row.

So carrots typically are, you know, they kind of they're kind of scattered around. And, and you just kind of like throw the seeds as you go out and implant them. Wherever they're last row crops, we actually like very used machines that are very, you know, particular about where they grow, where they are, there's less, there's less ambiguity, and the in what our place, they're very, very structured and very, very light, you know, every 10 inches, you know, just depending on particular crop, they're very, they're 10 inches of space from each other 30 inches, you know, from the next row, it's very, very prescribed, whereas non real crops are, you know, they kind of they kind of are planted in a row, if you will, but but but what, what, what typically happens is you plant a couple in a row, and whichever one is the best, the best vibrant one after a little bit of time, you kill the other ones, and then you keep the you keep the one that you want. And so river did that kind of work in agriculture in California. You can imagine there's lots of vegetables, fruits that are grown here and they've grown in that manner. You planted a couple of them, as you go down and the ones that's the most viable, that's the one you keep, you kill the rest.

Okay, and you kill the rest so that the one that looks most viable, or best at that time, has more resources available to it less to compete with in that spectrum is a better harvest. Maybe.

That's right, it's so better quality ones.That's, that's right. You know, you want to give that, you know, every, every person wants to bigger to see or strawberry, bigger, juicy carrot, whatever that is. And so, so and so that allows you to say, okay, hey, I planted these in a nice bed of a fertilized ground. And whichever ground is a is is a, you know, whichever one out of the of the three or four, that is more viable. That's the one that I keep, and then I kill the rest. That's right. So as you get more and more, you get more yield is what we typically call it, a grower typically, is concerned about the yield of the crop that they they are planning, and they want the most yield for for the amount of planning.

So they have a certain amount of space, and they basically want to get as much product, carrots or whatever it might be, as possible for that amount of land.

Okay, so with row row plants, row crops and non row crops like carrots, does the machinery change for because I'm imagining like a big tractor driving down tracks? And so I'm thinking for the non row crops, which you said are kind of in rows anyways, does is it like squishing a bunch of them? Is the row was I would imagine it's going kind of in the seams between the rows of plants.

And so yeah, so you can imagine it, there's like these little mounds of dirt you create, and even something depending on where you are in the country on the world. That can be different, but But essentially, you're maximizing the land space, and you're maximizing the ability of that particular seed to to be the best year. And so the transition. And so the transition is, is you don't you don't carve out so much land, because you don't know, when you know, particular crops like that are fruits and vegetables, you don't know which one will be the one that actually takes hold and is the champion, if you will, that part of the land to give you put three or four of them down to see which one is really taking the resources out of the ground. Yep, in a row crop, you know that, hey, if I plant the seed here, this probably gonna grow if I plan in perfect places, you know, soy bean, and the particular row crops that we're talking about soy bean, cotton corn, if you plant them there, they're probably going to be just fine. The thing that will be the resource, so the resource wouldn't be the resource like pool wouldn't be weeds, it would be the... Excuse me, I said that wrong. And a vegetable crop, you want to plant multiples, because you don't know what the crop itself will take? Yeah in the row crop, you know, the crop itself will take but then there's competition as weeds. And so now you have to kill the weeds and vegetables and fruits, you need to plant multiples, because that that we'll see may just not grow right on its own. So you can't just plant one and I know that that one's gonna, but you got to plant multiples. Because you know, out of the if I plant at least four, one or two of those guys are going to grow. Yeah, and then you and then you take the best out of the out of the group was in row crops, I plant a seed that seed is going to grow the only competition is weeds. Yeah. And, and so and so in that situation, you know, you're planning, you're planting seeds that are going to grow, you just now need to make sure that you you provided the best environment in order for it to grow.

Gotcha. So if I understand correctly, when you have the non row crops, you just have like a hump of soil that goes back in a row effectively denotes a non row crop and you just like throw seeds all over it. Right plants will come up anywhere in that big wide row and kill the ones that look less promising to save the ones that are more promising. Okay. And in the row ones, you just have a narrow row where the seat seat seats easy. And they Oh, yes. Right. And so in the non row ones you are killing the competing ones of the same species that are less viable. And row row ones you're more concerned about killing the weeds that are competing with the crop you're interested in. Is that correct?

Okay, so with going back a little bit, so with Blue River they were starting with these non row crops. And then feedback from customers was we are interested in row crops. Continuing from there, what happened?

So Blue River had to make a shift right that the, the technology, the actual even even the agro agronomical When I say agronomical, I'm talking about the Cool World, the word the the way that you know, you do something in agriculture,

Is it like economical but for agriculture?

agriculture, so So in this case, we have all the seeds that are out there, and we want to get rid of the ones that are not viable. So how did you do that, in a way an agricultural? And so and there's multiple ways you can play, you can spray herbicide, you can go in and dig up the ones that you didn't want, you know, or you can you can you can even apply her fertilizer, and that's what Blue River did they apply fertilizer. And what happens if you apply too much fertilizer to a plant? It'll explode it, you know, it'll just grow so fast that it can't exploded. Realize think about that is not that that fertilizer is in that soil. The other plants that makes the one that is viable, ie that one, how can can grow stronger, because now you have made a very fertile environment. The ones that were weak exploded, but this guy that is not weak, took that took that fertilization and use it and grew better.

So again, not knowing much about agriculture, why does it... so it explodes, it grows too fast, and then it can't support that or something? Why? Why? Why does it die? Because it grows too fast? Because it's just it's just the fertilizers a kind of Yeah, how to put that it is something that encourages the plant to grow very fast, but good plant has to be able to handle it

Does it grow to... is it like a startup in some sense, that takes venture capital funding, and it basically gets a huge investment, but if the money, so you grow rapidly, and then expenses are coming, so then the... Essentially, you're going too fast. And I would put a similar startup somewhere going too fast if you don't have the funding, but the but you don't have the infrastructure if you don't have the infrastructure to handle it. Right. So that viable crop, right, that one that is like it's gonna it's gonna sustain, that one has already has gained all the things necessary to reap the benefit of that fertilizer. These other ones are set in such a non capable state, that when you you feed them fertilizer, they, they grow so fast, but they're not capable, they're not ready for it. In other words,

So they grow before they're ready, and they overextend themselves. There you go, it's too much, this is too much ploom. And a guy,

I'm thinking of like Napolean now and having too big of an empire, and then it collapses. It collapse. And you could think about it as you could think about it, as I have too much resources that I cannot contain it, I cannot help hold it within my infrastructure. So I need to so I so what happens is it just explodes. Whereas the plants that are the plants that can handle that they take it in and they use it and grow, which is what to expect when you get you know, like, like an investment or an increase. And it is a ridiculous amount. Right? It is a it would be like, you know, you give somebody who's never had a business before in their life, you know, 30 million, right? Like, they don't they wouldn't know what to do with that property. Right? Yep.

Gotcha. It's like getting rich, quick kind of thing. And then it just goes away. Okay. They have no ability to deal with it. Right and no ability to deal with it. Yeah. And the thing about that is that that money doesn't go anywhere. That money is still out in the environment in the system. Yeah. And that and and that next crop that's next to those to those in those plants that get killed that way, reaps the benefit of it. So you're fertilizing. So it's kind of a two pronged process. You're gonna fertilize your soil anyway, because you want your plants to grow. But the ones that are not viable, you fertilize it to the point where they're going to die, but that overlying fertilizer still stays in the ground and allows the ones that are not going to die, to grow, to flourish to be better. You harvest those, that ground is still very fertile with all those resources that you come in are still there. You can go plant again next season, the grounds fertile, it grows again. You do this process again. So you're, you're kinda you know, it's kind of a cycle. fertilizing the ground but also getting rid of the the the weeds to the crops that are not going to they're not going to make it right if you give those guys an extra shot, so that they so that they they're not going to make it and then...

Is this row crops where it's weeds that are competing, or we speaking about non row crops, or both?

That was non row crops. So in row in a row crop, it's a different environment and that's what I was talking about agronomical tools, right. And so in a row crop. You know, fertilizing the weeds is not the way to go, right? You don't want the weeds to grow. You want to deplete the weeds, you want to show that was removed?

I see. So that was with the non row crop, what we're talking about with over fertilizing them. So you do it to the non row crop that are not promising. And thus, they grow too fast, then they explode and die. Not literally explode, probably. But for the row crop, you're saying now we have a different strategy?

That's right. And that's what I'm talking about the agronomic the strategy, right is that so and so in agriculture, in plants, you can imagine there's this a variety of plants and in the agronomic, the agricultural world is just as varied as humans, right, there's just life itself is just so varied. And it's even more so in the plant world. And so different strategies are needed for different types of plants that you're going. And in a row crop situation where you know, the seed is going to take, you know that when you plant that seed, it's going to come up and it's going to come out of the ground, it's just a matter of how much yield you're going to get from it. Your strategy there is different because now you're not competing with others, like seeds, you're completing what you're competing with, what we are completely different plants that are trying to suck the resources away from the plant, you're trying to, you know, the cash crop is what we call it. And, and so in that situation, you actually want to kill, you want to alleviate that leak, you want to alleviate those plants,

Alleviate is a funny word for that. It's the problem, you're alleviating the problem, and thus you're killing the plant. I see. It's killing the plant, it's killing the plant. It's making sure that that plant because you never really kill it. You never really kill it, right? Like like those, those,

But you remove it as a competitor for resources. So you think about plants, you think about life, life always wants to persist, it doesn't really die, right? It's hardy. That's absolutely true.

Only the cash crops die occasionally. We purposely we purposely harvest those, right, but the weeds are just there. And the interesting thing that happened, so just go back, you know, 30 years, they had what roundup they had they had, you know chemicals that could that would you know they have their crops genetically modified, you can spray the entire field, we call that broadcasting. So you spray the entire field with herbicide, the plants that you were trying to keep were genetically modified so that that herbicide did not kill that plant. But it killed everything that wasn't anything like that all the weeds that were in that plant that had a modification in it to that herbicide. And so all those plants and what you would see if you could go down a growth field and would be what we call clean, you would only see the crop that the girl was trying to grow, meaning there was nothing else but everything else was dirt in the field or soil. Only the crop was there. And that now was the American agriculture for 3040 years, you know, using these genetically modified plants. Okay, what happened was is over time he is talking about life and how hard it is you kill the just dude, you killed all the weeds that were susceptible to that herbicide. And the only ones that are persistent were ones that were able to fight through the herbicide. And so now they became they build up the herbicide resistance. Valerie's so now they are the old herbicide resistant plants are the only ones that are reproducing. So now when you reproduce you're only be producing herbicide resistant plants and that is causing a huge air a huge battle man gotcha because now you now the plants are being able to build up resistance faster than the engineers can build up new technologies to actually kill those those resistant plants. And so now now you're not you're having, you know, what we call unclean fields, fields where growers are spraying. They're planting crops and these herbicides are resisting it. And so now they have to come up with something different different chemicals different ability to get rid of the plants. And that's in the timeline of like the 30 years you said... So 30 years, 30 years we've been using the Roundup herbicide, we've been using herbicide, or I would say the last 10 years is where they really become maybe in the last five years is where they really become prevalent herbicide resistant weeds, to the point where they're, they're, they're starting to, like, you know, hey, the strategies that growers use, and there's multiples, as herbicide, there's a there's a will be called a field fighting you call when you kind of just trim up the dirt in the row to kill some of the weeds. There's multiple techniques, but but like I said, you know, alleviate the problem, you want to give the crop that you're trying to grow and get killed from the one that you're trying to sell the best opportunity to win in the, you know, in the competing of resources that are in the ground. So you want that crop to get as much of the resources as you can, and to provide, you know, the weeds and other things, less resources so that they die off?

This may be a silly question, but why not just make an incredibly resource dense field and have a bunch of weeds but also have your cash crop? Because the weeds the weeds are. The weeds are. They're called weeds, right? They're just plants. Yeah, but we call them we undesirable not cash plants. And the reason why is because they they're very effective and sucking the resource in that you won't your cash crop will not crop, they will take over.

Starving it of resources and they grow way quicker.

Their genetics are so that they will take over. And those viable those viable crops that you really want that you want to keep, you know, which was you know, when you think about, you just think about society, and you think about, you know, just biology again, in the world, you know, things that are valuable to humans, particularly the humans are things that are scarce. So things like you know, cotton soybeans, plants that you could eat plants that you could, you could read, you know, something that was due Sunday, they weren't just plentiful everywhere, and everybody could just have it and it was great. In order to make that, especially for the population of the world today, you have to very strategically go and grow it. Right. That's kind of how the world got as populated as it as it has, is because we no longer just kind of go hunt and gather and just go find where they're where the right stuff is off the tree as it grew naturally. Because in order for it to grow naturally, you would have to get in an environment naturally, that just didn't have enough competition with all the other farmers in the area, so that that thing could grow. We're very specifically saying, Hey, we like apples, we like strawberries, we like

So we have to definitely beat off the fauna that would that naturally just grows, that naturally will just take over. It is and isn't, you know, necessarily beneficial to us. It's right, it's plant life, but we're trying to specifically grow plant life that's beneficial to us. And all the other plant life around is like, hey, I want to use this. I want to take this spot up. Yeah, you're and they don't use much energy, right, that the reason why those things are valuable is because fruits and things like that need a lot of energy from the ground. And so, so it's easy for these other things that don't bear fruit that don't give us any benefit to grow and rapidly take over the land because they don't, they don't need a lot of energy because they don't provide anything of value to us. Fruits and you know...

Stuff that we can harvest from them. You can think of the things that that humans care about us the fruit, right. And, and those things. Take energy. The things that we don't provide fruit, necessarily don't don't take a lot of energy. And they they're prevalent. They're everywhere. Yeah.

And by fruit you mean anything we can harvest. Right? So like, I don't know, cotton? Like the seed probably is not a fruit. But you mean that as in something we can harvest and then use? Is it correct? Yeah, absolutely. So the bulk..

Gotcha. Gotcha. Okay. And then so what was the strategy for row crops? So the strategy here for row crops is to, it's a cool thing that Blue River did, is because you're talking about agriculture, you're talking about a customer in a market that's been around since the beginning of the United States, right? It's this goes way back. John Deere has been a company for hundreds of years in the US. That's awesome. And so the agriculture is is a mainstay in the United States. And that customer in that in that market is not very open to drastic changes. They've been doing things the way they've been doing it for the past 20 years. You're not just going to come in here and say, Hey, I got the spaceship or I got this cool technology, they're just gonna make your you know, they're very adverse to those kinds of things because they know what they've been doing. They've been doing that for for, you know, for a long time. And they're very adverse to very crazy out of the box changes. And so what we did was okay, you already use herbicide, you're already spraying your entire field, with, you're broadcasting your entire field with herbicide to cover every anchor, right. So if you got 1000 acre field, doing spraying the entire 1000 acres with herbicide, even though there's only beads in certain pockets of that field, so you just bring everything but Blue River did was with the court and we did have this novel was a you know, add a camera to that, to that a sprayer, the same spray, you always been using the same technology, you used to use any like pointed out things, we added a camera, the camera can look at the ground, it can tell where the weed exactly located, and then we spray onto the weed. Instead of spraying the entire field, you only spray where the weed is.

So now you only kill the herbicide you only spray where we need to spray the herbicide. And you can imagine like you're reducing the cost of the grower, you're reducing the the herbicide that's in the ground, you know, causing environmental impact. And, and so and so now in the give the girl the opportunity to use the same technology that he's used to using, but being more effective and more efficient, and you give them the opportunity. I was talking about the herbicide resistant weeds, you give them the opportunity to use different chemicals. Yeah, that's right. That's right. Because before you sprayed everything, so you have to use an editing modified herbicide in crops, because you want to be able to spray everything, you don't want the crop to die. Now if I can just tell you, I can just hit just the wheat and not everything else. Now you can use different chemicals, different herbicides to target the week. There you go.

Now do you need to use genetically modified crops for this? Because you're killing things in the same field with some sort of herbicide? Or like, do you want them to be slightly resistant to it? So that something in their system is not like say you spray this one weed in this field of cash crop? And does that impact negatively the cash crop that's around it? Or is it a small enough and precise enough dose? That it just kills the weed and doesn't damage the crops? Or is it a trade off between the two. So nothing's ever perfect, right? So there's It depends, depends on the cash crop, you're using some cash crops, you, you want to be able to, you want to be able to genetically modify them so that you can spray the crop as well as the week. And then depending on depending on the cash crop itself, you don't want to spray it at all, you want to spray justly. And so and you think about you know, the situations and the variances of fields, the variances of crop types, the variances of wheat, you know, you have all those different things, and depending on the situation, you want to apply a system that works for your situation, right. And so the focus of Blue River, and John Deere in overall is to give the customer the ability to tailor their system to their situation. And give them a system that that effectively will work for them, given the situation. So some situations you can genetically modify, spray, just the wheat and a little bit of the crop and it doesn't hurt some situations you want to be able to spray, you want to target test that weed only and you don't want to spray the crop at all.

One thing just to clear up: Would you talk a bit about the relationship between Blue River and John Deere. Because we mentioned it but just to like explicitly talk about how they are related. So Blue River was around right there was a startup company I talked to you about the founders. Yeah. And and eventually they you know, not just John Deere, but other agricultural companies were interested in purchasing Blue River and I think John Deere won that war, the war if you will, and

It's nice to have a bidding war. Sure. So the use of the technology and the viableness, and the need, I think is real, like it is absolutely 100% a real problem that's happening across the world. And it needs to be tackled it like it has to be tackled. John Deere was was one of the first people to say hey, this is a you know, not just not even just this particular problem with the herbicide resistance,

Yeah, yeah. Just, you know, applying artificial intelligence and in high tech, high tech, you know, things into the agriculture, agricultural world, that now that's a new thing, right like that these these, these companies, you know, John Deere particularly, have been operating and using, you know, machines for for, you know, hundreds of years. And they have grown a Technic technological advances, but, you know, they haven't been implementing, you know, artificial intelligence and, you know, high definition cameras and high, you know, a very high power computers. Yeah. And so, so making the next step, I think John Deere thinking about making the next step, and where does agriculture have to grow, you know, and for the future, to continue to sustain and to continue to maintain, you know, for the population that we have on this earth, that these technologies need to be in, you know, integrated into the agricultural industry, and saw Blue River as you know, what the way that we want to tackle the problem using AI, using very high performance computers, using, you know, advanced technologies, you know, they in robotics, they saw that this was a nice day, this was a nice, a nice integration into their company, gotcha them to give them the ability to, you know, to use robotics to robota size, if you will, their machinery that they had already had, that they already built, they already make.

Gotcha. So we then see, we're spraying herbicide on the weeds in a cash crop, a row cash crop field, is this kind of the, I guess, where does Blue River go from there before being acquired by John Deere? Or is that is that kind of the core competence of the company? So So we essentially, you know, the mentality is give me up to give the, the CRO or the plant by plant knowledge, so that they can make a smart decision. Right, so we have cameras, we have artificial intelligence, we have the ability to make real time decisions using robotics, right? So we use robotics to be able to say, hey, real time, this is what this plant needs. And that was Blue River, and still is, it's blue rivers, you know, driving force, you know, you can look at the plant real time and decide right there. What does this guy need. And that is, you know, when you, when you think about any, any person who's in agriculture, or, you know, planning biology, they will tell you, the plants that grow the best are the ones that are taken care of individually, this plant needs this, this one needs a little bit of pruning, this one needs some fertilizer, this one needs, he has weeds growing around it. So Blue River was allowing you to do that and, you know, skill level, you know, not not your pot in your house, you know, where you kind of hand doing in the field, in the field and 1000s of acres with a machine, you know, that you can just kind of let it do its thing. And you as the grower don't have to sit there and do that.

And is it only is it only spraying plants? Is it also trimming or any of these other things that you've mentioned. So you can you can imagine the technology can be advanced, if the original, you know, this initial, like induction into the market is spraying plants for herbicides, because that is a very prevalent problem. And it's something that Blue River was already doing. But you can imagine.

It's cool, because you can retrofit existing systems. So that's very nice.

Retrofitting existing systems, developing new systems that are very, you know, optimized. Yeah. Then and then extending the see and sense technology, kind of like, you know, we talked about this a little bit, you know, seeing and think and actuate, your podcast. That's exactly what Blue River, you know, long term, you know, overall, that's the goal is to be able to sense the environment, do something smart, and then actually do something smart to the ground or do some smart and get world sense. This is exactly what your podcast says and think about it, think about agriculture, all the different things that you have to do, you have to plant the seed, you have to take care of the seed when it's down in the ground at the fertilizer, you have to weed it, and then you have to harvest it, and then you take it from harvest and you have to get it out to the market so that they can be sold. So there's just a lot as there's a wealth of, of tech, you know, this, the agriculture industry is just I won't say behind but just has been doing things a certain way for a long time and it's a fertile ground. For new technology for New for new ways to go about to bring that that sense. And they can actually wait, you know, paradigm into that world. There's, there's just oh my god, it's just so many different ways and opportunities. Yeah, that's absolutely right, John, John Deere and Uber are being acquired by John Deere to get us right in the forefront of it. Because, you know, that's the one of the biggest agriculture companies in the world. And, and so and they do all that they do everything, you know, from planting, to harvesting to, you know, to tilling the ground to getting it prepared, and as well as other industries construction. Forestry.

So what do you what do you mean? Oh, are you saying John Deere does these?

Yes, yes, yes, John Deere. And so Blue River, you know, bringing in the sense and think kind of a mind frame, into an industry and until a company that already has its footprint across our world and all established? That's right, all.

And they're, they're I mean, they're domain experts at the actual markets that they're servicing. So agriculture, and then other ones you mentioned. So then partnering with them, where you guys are being you're helping on the sensing and thinking side, and then they can kind of use their existing systems are retrofit your systems into it, or even add your systems design your systems Integrate them, exactly. Really great, really at the foundational level, right? Let's develop new systems. But let's rethink how we do this. We know how we do it already. Right? But we do it, we did it already with previous Well, you know, what previous experience, now let's bring in this new technology, and let's rethink how we could do it. You know, if we, if we bring in, you know, different sensors, cameras, lasers, and sensors, you know, smart algorithms, and different, you know, different ways to actuate in different ways to go about doing something. And then now, you know, and now change the future, right? Like, you know, you could just think about all the different ways, you know, that AI Robotics has changed the world, and it's gonna continue to check the world, you know, before you have to have a human in the loop, right? Yeah, human beings, humans just do it fast, literally, when it's asked. And now, you know, you can provide a smarter machine that can do that can do this task.

Or one person can do a bunch of fields and effectively have a bunch of like, robot workers on the tractor doing the hard manual things. Absolutely. Very focused, intensive. Absolutely. Let's see. And then. So with Blue River? What? Wait, so I'm just curious, I guess, how was it? You were around? When John Deere purchased? Blue River? Correct? That's correct. Yes. How, how have things felt to you, as things have changed with that? Like, how is the focus changed? Or what kind of problems are you guys working on? What's interesting about being acquired, and the process of that?

I think it's, it's very, like, for me, personally, it's been very interesting just to see a company transition from you know, very, you know, startup, you know, environment, you know, get out there and fire in the belly, we got to make something we got to continue to, you know, develop this product that people are interested in, grow into now, you've been acquired by a company, that is totally has the resources to make sure that you grow, but we need to grow it in a direction that is actually going to be viable, and actually going to be, you know, a product is actually going to be put out to market, and customers are going to receive and use, and, and having all of that expertise and that knowledge of all those years of John Deere manufacturing, agricultural sprayers, and any piece of agriculture milled with the high technology. And you could you could already think that there's going to be a lot of challenges, right? That's a two, not necessarily two different worlds. But you know, you know, kind of different environments kind of have to mesh together. And I praise John Deere, for giving us the ability to kind of keep our autonomy as far as developing the technology and going as fast as we can to continue to develop it and grow it and, and make changes as we need. But keep us kind of, you know, I would call it. What do you call that leadership and management? You know, John Deere has this division, right? This is where this thing needs to get funneled into. You guys have the ability to do whatever you think needs to be able needs to be done in order to accomplish this, this goal here and we're trying to accomplish and the goal is big enough and broad enough, right? That it encompasses all the things that they're doing, right? It compasses, not just the sprayer part, which is just as one part that we're working on. And probably the first thing that will roll out, just because we're Were Blue River was the most mature in the area. Yeah, but but we could, but we're already developing other areas. And so And so John Deere just has a great vision has the great wherewithal. And I think because they have acquired companies that just have all the experience, are they the company's been around for years, they know, you know, the ability and the way that they they went about the acquisition after it happened, and the way that they're slowly bringing a sense of their, into their environment, and so their world, but still allowing us to be Blue River, I think is great. And I think it says kudos to them. Because they could have totally just came in and say, Hey, boom, we bought everything we own everything you got, they're gonna do his like how he said, and they didn't do that. They, they, they, they said, hey, they appreciate you guys have something special here going on. We don't want to affect that. Want to allow you guys to continue to do that. But we want you to do it in this direction. Yeah, in any directions aligned, which I think is another, you know, smart thing on them. And I think what Blue River as well, to reason why we got acquired by John Deere is because the directions aligned very well. So and so it was like, Yeah, this is gonna work. Right. This is gonna work, you know, kudos to our CEO, for you know, navigate that because Yeah, absolutely. Absolutely. You know, making those making those things happen. And then he's still around, right, like our CEO still here or is still doing job. You're still managing he still going? You know, very important things for us. Yeah. And, yeah, we can say that this is working out pretty pretty nicely. I

Hell yeah, so what was the timeline? So founding of Blue River acquired by John Deere... We got founded 10 years ago, that's, what this is 21, so 2011. Yeah. So and then. And then, we got acquired by Deere in 2017. So that's what six years after the founding, they got acquired by dear God. Yeah. And so here we are 2021, four years after acquisition. Still, still going well, and probably going to bring the first product to market. Two years from now.

Two years? Can you talk a bit about that? Or is that a secret? I can say that it's coming. That's about it. I mean, not any details, but it's, yeah, the first, the first product will be is coming soon.

Nice. That'll be exciting. So then, can you tell me a bit about your experience? So you, you Quality Manager, Quality Assurance Manager? Can you tell me a bit about what that means? And I mean, I've been in academia most of my time, so we don't really have this, would you tell me a bit about your job as a quality manager? Absolutely. So, so Quality Manager is, you know, this is an essential part of the business is to make sure that the customer receives, you know, value in that product, and that the product that they get is what they expect, not only what they expect, hopefully it sees expectations, but But you know, first and fundamentally is what they expect. And, and so it's making sure you know, I think about it this way, you can have the startup mind frame, you know, and just really just gone, you read this, move fast and break things with us break things, you put something out there, but there's all kinds of bugs with it. And depending on your market, and depending on the customer, you think you think about your, you think about these phones, you know, 20 years ago, and they and they had all kinds of bugs and issues with them. But the customer you were dealing with was okay with that right? There early adapters are okay, yeah, I can I can I oh yeah, I gotta do this thing, or I got to press it two times this way and do this in that way. And then it finally works. Right? You know, so that that was fine for that environment. But for our particular customer, you know, agriculture customer, they want to bring in John Deere, you know, name recognition. You know, it just works. John Deere has the reputation in the in the, you know, the cloud that when I get a bite when I buy a John Deere, it works out of the box, I turn the key, it does what I wanted to do. And if anything is wrong, they fix it Johnny on the spot, no big deal. High quality high, you know, I pay a premium, because it is the high quality product. It's the best, right? It's one of its, I would say it's the best, right? It's the best that you're going to get. And so how do we maintain that right? How do you know John Deere has maintained that over years, over centuries? How do we maintain that and so my job is to allow blu rays For developers, the ability to continue to grow to continually to develop this product and continue to iterate and make it better, you know, the Bay Area Silicon Valley, quickly area area can fail. But when we release something out, or release something out to the to the, to the customer, it is it has that John Deere quality it has that this is going to work, it is robust, it is going to do what you guys told me it was going to do. So I'm the kind of the gatekeeper. I'm like, yeah, go develop all that cool stuff, go get in the lab, go write your code. Let's get out here in the field. Let's go test it. Let's go. Let's come up with all the cool things that we can. But this is what we can release. This is what we can, this is what we can give out. Because it's ready. If it ain't ready yet, you know, hey, it's not ready yet, guys, we gotta go back rehash. We, you know, reiterate. You know, and so and so I want to allow those guys the ability to continue to reiterate and continue to develop product, you know, features and things, but only allow John Deere quality features got ready to go. And so that's, that's my that's my cool, cool. Yeah.

So does it what does it look like in a little more detail? Is it like when I think of quality assurance engineers, I mean, I'm at open robotics, and we have a lot of software. So quality assurance for us means like, writing software tests, or making sure we have certain amount of code coverage with our tests or whatever, these kinds of things, it sounds like your work is a good bit more applied as in like, go in the field. Can you just tell me a bit more about what it involves?

It's that right? Is that and but But you said you said exactly. It's definitely you know, unit test, you know, code coverage of things in the software stack, right. But it's not just a software piece of equipment, it's a robotic piece of equipment. Yeah, and so and so it has to actually perform out in the field. So there's more things than just does the algorithm behave the way we expect, but just the mechanical devices to behave the way we expect, can can the cameras and the wires and the things hold up to the environment that we're going to throw them in. And then again, I said earlier about, we gave the customer the ability to use a platform that they were already used to using. And they're already used to use it in a certain way. So they already used to have in this sprayer that I parked out on the corner over here, and it rains on it, and it it snows on it, if depending on what part of the country you're in, you got it that the environment you're paying to it, but I can turn that key next season. And it goes and it does its job. So making sure that all of the pieces and all the all of the parts come together the integration, in a way in a way that that when that customer is actually my North Star is customer turns, the key thing does what it's supposed to do, or at least tells you hey, you need to go do something, you need to go do something right, you know, your camera cables, have fried over this past season, go replace them, you know, and so making sure that all of those things are there, the quality is there in and of itself. And when it isn't there, it is very obvious to the customer, what it is they need to go through to fix it. And so it's it's you know, it's unit testing, it's called coverage, like a software company that has those features. But it's more customer focused, it's more, it's more customer focused. So we definitely have to go out and test it on the machine and give it real machine time. Yes, see, to see what the customer is going to experience and give them the best experience we can get.

Okay, I'm curious about the hardware side of this. How do you deal with software, if I write tests to constrain my code to make sure that it's working, the tests fail if the code is not working? And then I just run these frequently. And I can tell if my code is working over time. What do you do? So that that, to me is clear for software? How do you test hardware to make sure that or even like hardware and software and their integration together? How do you guys do this? Yeah, no, I'm sure. I've seen like with Boston Dynamics, or some robotics company, they'll just have their robot do the paces. So they'll have it run and move continuously for a week or something like this. But can you speak a little bit about this? How do you...

I mean, I think it's essentially the same similar things. I mean, I won't get into the details of exactly how, but it is something along those lines, right, we run we run that system through the paces, if you will, I can't give you all the details, because that is it is essentially the John Deere name, their reliability and The and the you know, the the the confidence you get from saying that I bought a John Deere and I know this thing is gonna work. This is part of their secret sauce as part of their, you know, this is how we this is how we got to be who we are. And the I would say I will say this, the nice thing about being acquired by them is that we were able to be brought in. And so that process of how they ensure reliability and quality to the customer. And we could add that in and that and I think they you know, a lot of startups, not every startup, you know, there's always gonna be great startups, they're always gonna, you know, hurry up and fail, and then figure out the quality. To do that, and they're gonna raise it, yeah, they're gonna do it, right, they're gonna figure out oh, that didn't work, change it up real quick. That works now change it up. And depending on the customer base, you have a lot of units, you have a lot of room to do that, especially if it's new technology that's never been, you know, out there, and your customer base is like open to it. You know, our customer base isn't so much open to things that don't work, it has to work. And so John Deere is very, you know, keen on that. And, and so all the 50 did a good job of bringing us in allowing a startup to see how did you change transform from making this cool technology that works? You know, at first, you know, we talked about you could talk about when you develop something, you know, is that 95% edge cases are the maker breakers, right? You can make something that works 80 85% of the time, the 9095, you know, the 9798, those ones are your problems that take the most time, right. So those edge cases, and John Deere has a has a method for that. And so and so we and so we work, right, we continue to work, and we continue to develop the product. I can't get into any of the details other than that, like you said very clearly is that you continue to test that system and method, you have a method of doing it. So that you know that when the time you release it. It is ready. It's ready for the customer is ready to accept it is not is a perfect? No. It's that there's no perfect no no perfection. Oh, of course. Yeah. But you have all you have all you have all the big things out. And then the rest of it is there's a system to have the rest.

So is there. So it sounds to me, like a lot of your job might involve figure making decisions about what can and cannot be included? Can you talk a little bit to this, I'm sure it has something to do with the John Deere process. But also, it's a value judgment. On your side, I imagine where you're going, we need it to be good for this temperature range, or whatever it is. That's right, that, you know, it's essential, right? Like, there's specifications, and there's, there's certain things that are, you know, go no, go, I try to be very clear on that. And then, and the other part of my job is when that is included, define some things are just clear. This is good enough, and this isn't. And then there's some things that are like, Okay, this is a gray area, you know, we don't have a really clearly defined.

Can you give me an example? So, so, let's think about the technology, right? Think about artificial intelligence, artificial intelligence, you know, it's probability, you know, so you have these, like, you know, these algorithms and these metrics that say, okay, hey, we're about this percentage, good, on on our detection of...

So maybe like, I'm 95% confident that this is a tomato plant. There you go. 95%, is 95 going to be good enough? Do we really need to be 97 or is that is that, where was that number? And that's hard to say? And you think about the world, right? This these machines can be anywhere in the world, that various changes depending on where you are. So you're over here and this part of the country or this part of the world, that's good enough and other parts of the country of the world or particular actually even a particular actual grower, that's not good enough, I need this thing to be this well, or otherwise, I'll just go I'll just go by the other thing.

So work. So I'm thinking like, okay, considering you guys are in like all over the world, you have machines all over the world. I'm imagining knowing that if you're using some sort of like probabilistic machine learning AI thing for identifying plants or weeds, there's probably a good amount of variation in We there's a good amount of variation in the type of lighting you're get during the day, there's a good variation in the amount, or like the color of the soil, or the makeup of the soil, or anything like this. And this would probably make it depending on the method you're using. A good bit more difficult to come up with metrics, I guess, yeah, this would be a good gray area examples, right? That's exactly why I brought it up. Because, okay, it depends, right? It depends on, like, all those things you said, right. And probably oil more than environment, the lighting condition and what condition you're in. And in, I think the moisture important, the important piece that I talked about what quality is, is under letting the customer know, how, what quality you're getting. Right. And so being able to communicate that, I think that's an key, I think that will be key in the future of robotics, as well, as we continue to develop the technology. There will be situations where there's an expectation, and now we cannot guarantee you what you'd expect. And so you know, we tell you, Hey, you're in a situation now that we're using fuzzy probability mathematics. And yeah, your numbers are are starting to below come below a threshold, yet. So now it is up to you as the union user to say, Can I keep going this way? Or? No? I think that is, I think that's I think that's going to be I think that's going to always be the case, for a long time in robotics. And so we can get all the edge cases out of out of a particular application. And well, how many robotic applications do we actually have in the world? There's not a lot of. And so as we develop these new novel ones, though, they'll need time...

...to mature even more and become more reliable. Okay, does it mean, you go and test the robot system in a similar environment to what the customer is, and then measure how effective it is, and then report that to the customer? I assume it's not like a real time, this is how well, we're doing. system when you're talking about telling the customers quality information. So it's both it's, it's, it's actually some there's some real time to it right? There has to be there has to be in particular, it's a robotic system, you have to let the customer know, like, as real time as you can, that we're not getting like you're not behaving, you're not behaving the way that you want to behave. Yeah, so you got to be able to give them that you have to be able to give them that.

Okay, I guess I'm a little confused, because I'm imagining like, a robotic system going and spraying plants. And maybe it says, we hit 80% of the weeds. But the thing is, at the time, it thought every one of which it sprayed was a weed. So 20% might have been the cash crop or something like this. So the secret sauce, I can't give you but but what I can tell you is that you have to be able to allow the customer to know that your system isn't behaving the way you expect it to.

Oh, right. Maybe it's just being aware of how confident you are about the things you're doing and reporting that.

Exactly. So So yeah, exactly. Right. Like,

So it's confidence based, not outcome, you don't have any real labels for what you're doing. Gotcha. Well, so so. So it's, it's so yeah, so it depends, right? It's a combination, I can't get into all the details yet. But there is a I will put it like this, there's both components, there is a real world component, that is real time. And there's also a a introspected after the fact, you know, but you know that you can go back and look at and you can and then you can combined. Right? So so I'll put it to you like this, if you ran over a field once, if you've already kind of been through an area once. And then you go back through it again, you know that that provides you ability to make different kinds of decisions.

You get to see if the weeds were all killed, or if there's a bunch of them still or... Where the weeds are? What, what's the density of them, you know, how the machines behaving at the time when you go through that part of the field? There's lots of different things.

Cool, okay, I didn't consider that kind of temporal part of this. There's lots of different things that we definitely do both. I would say we do both something real time because you want to allow the customer to know right away that hey, you're not in a great situation. And but then there's also like, what is the situation exactly in that area? What is actually going on? Why isn't that a great situation? And then how do we correct for it? How do we make it better? How do we improve upon it? And so yeah, providing the customer with the ability to do that. And so those are two, I would say, those are two things that I constantly have to make sure that we're doing. And you know, it's cool, because we're developing both, right, we're developing both of those infrastructures. Now real time, like, you know, and so when that product gets released, you don't present the customer ability to do both, he can say, hey, right now, this is what's going on, what do I do? He can make a decision, or she can make a decision on what do I want to do now? After the fact that can come back and say, hey, that's what happened at that time? How do the next time we go through that situation again, time better, gotten better prepared and better to make a better decision? You know, what new information? And then and then you know, that, that turns on another, you know, opportunity or ability in ways to, to tackle it, you know, versus the real time real time you're making a decision right now. You know, planning ahead of time is always better, right? If you can plan ahead for that's always better.

Now, would you talk a little bit about doing work in the field for this kind of thing, or like, getting out in the field and testing your systems? Just I think that's a very interesting part of what it sounds like your job involves?

Yeah, absolutely. So we had to get out in the field. And I got a team of people, you know, and also in coordination, not with just my QA team, but the the, the field team, that's always out on the machine is always out there. And being able to, to understand how the machine actually behaves in the field. And so, you know, that it's not the same thing as, like an iPhone, or, you know, a piece of software that's presented in a web server, it's an actual piece of equipment that's put in the field, and another customer uses it, you know, puts their hands on it, and, and actually gets out there and use Google ads. So it's very valuable to me to get the customer's input the customer's point of view, the novel use cases. So putting out something in the field with a customer, you know, you have in your mind as an engineer and the engineering team, this is how we think you should be using our product. And the customer will just say, Oh, I can do this, this and this, and totally something you didn't even think about. You don't even think oh, wow, well, I didn't know that you're going to go through this. Because, you know, we didn't design it for that. But it can totally be done in that way. And, and so that's always cool to get their their point of view to get their use cases, because then we can start to think about, oh, wow, this could be used this way. We didn't even think about that. But let's make sure that way works. Let's make sure that that way works, because now we know, right? They want to use it that way. Yeah, they want to use it that way. I'm putting it out in as many I think any company wants to put their product and all the as many environments and use cases as they can to get the 97 98%. Because you don't know, right? Somebody is using it this way. You have your describe where you want to use it. And you don't know, you know what those edge cases are in your software stack, your mechanicals that, you know, of what's what what did we break down at? Were we not good enough that, you know, and, and so if you don't get out there, and you don't give yourself the ability to to, to do it in as many different environments as you can, you're going to find an environment that you didn't prepare for that you're not capable of handling. And you just have to figure that out later. So we're, we're keen on, you know, and the nice thing about being acquired by John Deere is we can go a lot of places across the country and across the world to make sure we're getting the big slots of use cases.

Gotcha. Cool. Let's see, blowing up a little bit. Or going up a few levels. What do you think? Where do you think robotics and agriculture is headed?

I think you know, you know, robotics in general is the future of all industries. Agriculture is going to totally, you know, 20 years from now it's going to dominate. There's going to there's going to be you know, these machines will be automated, and, and, you know, self sufficient, you know, 2030 years from now, or a girl will be able to say, hey, I want this to happen on my land, and it will happen and it will be very little human input. You know, there'll be there'll be some spots for human input. But you know, unless someone from Congress or somebody, you know, artificially removes the stress that is going to grow. I mean, it's the only way for these things to grow, right? Like there's, there's ton of machines as Thailand, we have to feed in a growing population, bunch of people. Yeah, we have to do that efficiently as we possibly can. We have to take in as much information as we can about the ground, what's going on on that ground? What do we need to do? How do we get the most out of it, and then get it ready, again, season after season, year after year, and we've been doing that as humans, but a lot of human input. And I mean, that works. But in the future, that's gonna, it's gonna be less human input is going to be a systems input, because we're finding that the system, you know, systems can can remove that human input into a better and more efficient. And so FFI anything, you know, fair history of history says anything, when we find more efficient and more effective ways to do something we do with that way.

Oh, yeah, for sure. We're, I'm curious to where you get the number of 20 to 30 years? It's just the technology growth? Right. Like, it's the, you know, that could be off, right. I'm giving myself buffer there.

For sure, I'm just curious about order of magnitude. It's the technology? It's the ability, you know, so AI is very, you know, I won't say it's infant. But applying it into real world problems is still fairly new. Yeah, there's some there's some areas, but we're talking about agriculture, agriculture, I think it's still fairly new, like we were, I think we're at the cutting edge. I think blueberries probably ought to be right, right at right at front, right. And so and so what and where does it go? And what are the new things that we're going to learn? And how are we going to apply the AI and the machine learning to this environment? You know, along with, you know, the cameras that can can withstand the rush the harsh environment, but still give you all the information? Yet all the like you talked about a little bit earlier, the cloud changing the lighting changes the all the things? Yeah, all the things so much variance, so much variance. So you, you put your dino, you know, it's nice, very controlled environment, you're in this very real world natural environment. How, you know, cameras, just one thing, you could do this, all these other things, right, it's all the other sensors, and then the machine and the machinery and what you're trying to actually do with that machine. Can those you know, the human hand, right is a, you know, a very complex, very dynamic piece of machine, that we still have a very hard time. robota sizing? Yeah, awesome. Dynamics is pretty cool. Right? I think they got some pretty nice dynamics, but I don't think they have a nice hand.

Yeah. Focusing more on local notion, the local most universal robots or right hand robotics or some of these other gripping, I don't know.

Those gripping technologies, right. In the garden. It's been, yeah, it's really simple. It's for it's for the application, right? It's for the application, right? It's like, okay, if I want to grab a thing I can do I just need a thing that does this. So that that mimics some parts of the hand. Yeah, but the hand can do so many different things. And so, you know, just, you know, how do we do you know, so that's the mechanical, more complex, the categorical,

More sophisticated in the mechanical part? Because yeah, you're right, right now we have very, it's like most very fine manipulation. With, in an environment where you're, like, in a highly unstructured environment doesn't seem to be a thing yet. There's lots of there's lots of dynamics right? And when you get when you get out to the real world talk about agriculture, you know, that this this this harvesting companies now I think they got robotic, like, they got he's got suction things, but he's still suck to suck the thing with a vacuum. Right? And so that I can kind of work in if you can manipulate it, you know, and you suck it...

Little things for picking strawberries. These kinds of things. I think I saw this at the Queensland University of Technology in Australia. You can get in there, and you can kind of grab it, you know, and that works in those environments. But you know, there's so many diverse environments.

Yeah, it's hard and not very general. Yeah. At this point, I believe I mean, I don't really much. I think I think initially there's going to be people who are I got a strawberry grabber, I got an apple grabber, and then somebody is going to make the universal So I grab stuff, yeah, whatever it is, whatever it is, you want me to go grab, I go grab it, right? So the mechanics have to get to that point where I can just grab anything, you know, and the human hand can do that. So I would imagine this, because how our brain thinks that we make stuff that is, you know, but then again, with AI, they might say, Oh, no, your hand is not the most optimal grabbing device, that's other device that you can grab, it's completely different than everything else. So that's the other cool thing that AI brings is the ability to rethink about our own, you know, designs, we come up with these designs, and it's like, oh, that's not the most optimal way to build the structure. If you build the structure with these very like, to us, they're not submitted, we typically build symmetrically and you know, the forces are very nicely applied across these symmetrical, you know, trusses, good things. And AI can say, oh, no, if you put this really long one here, and this one here, it has all the same constraints set as your Is there a symmetrical thing that you've been developing for for hundreds of years? Are you

Are you familiar with much work? I'm not? But are you familiar with much work where it's like design related things that use some sort of maybe an evolutionary strategy or something in AI? Is this what you're about? Yeah, we definitely. So we do some we do some developing Blue River, right? It's not? It's not, it's not, you know, across the board type of thing, but we do some of our own of that, right. And we do sort of our own, how can AI affect the design? That's what we do some of that.

Something or, like, the length of a part of it. No, I think I think it is just like, you know, you fundamentally give it a problem. And, and let it do its thing. It's actually like rocket science. We got some of our guys are like, you know, smartest guys ever. And we got some super smart guys here. Oh, my dad, and, and, you know, some of those guys just go out and say, hey, I want to do this thing here and give it a problem. I'm gonna let the I'm gonna let AI and the algorithms and the techniques come up with fine. Yeah, there you go. There you go.

Hmm. I wonder, I wonder if there's difficulties there. Because of I assume, well, I'm pretty sure you would have to do it in simulation. And in simulation physics doesn't really match up perfectly with our Oh, no, that. Yeah, yeah. So like, the attack is very hard to simulate.

So you gotta you got to get the you got to get the test stuff up, right, you got to build something, you got to, you know, prototype it get a texture rig up, see what you can learn, you know, in river river. And I think I think most companies are all about that, right? And every engineer would love doing that. Right? Oh, hey, there's this new thing. We got to develop this new test rig that that's saying, Yeah, let's figure out how it works. And totally, we, you know, that's, that's part of the that's part of the cool part of Blue River is that, you know, depending on what project you're working on, there's still that goes on, right? There's still that, hey, I got an idea. I don't know where this is gonna go yet. But can I get a little bit of budget to build a little something, and let's see what we get. Cool. And if, and if you get down, we started to see promise. that'll continue to get fans that will continue to get like, Okay, we'll keep developing. Keep keep looking at that. Keep figuring that out. Keep coming up with that. And I think I think all companies at some point, you know, I think they do they have to you have to do that. Right? If you want to, if you want to, you know how to have a John Deere, you know, stick around as long? Because we have to have somebody out there. Yay. Yeah, I want to figure out how to do something that is totally no one else is thinking about. Right. Yeah. Totally put a little bit of investment in that. And, my God, what can come out with we've come up with some cool stuff is, you know, just scratching the surface. I think we got some of the right people, you know, kind of like leading the direction there to continue us to grow there. Yeah, I think I think yeah, I think that that, and you talked a little bit earlier about, you know, the kids and the growing of the population so that they are capable of bringing those ideas seed.

Yeah, actually, that's that's kind of a good segue into what I would like to ask you about, which is I mean, so you have kids, and you've been encouraging them to program would you tell me a bit about this? Yes, so So I got I got two daughters, my oldest, now she's a nurse. And so she was a little bit behind the power curve when I, when I got into engineering, but but but I did explain to her that, you know, robotics, you know, computing and, you know, these actual devices that she uses as a nurse are going to converge, right, and that device will do a lot of things for you that you're doing right now manually as a nurse, and so that she should pay attention to it. Right. And so she, so she, you see has Riley she's been, you know, keeping up with, you know, a little bit of the technology, magazines, you know, around around the medical field, and the new devices that are coming and even like, you know, I talked to her about months ago, she was talking to a doctor about these new these new What do you call them, you know, so you can a doctor, they put that little thing on your your finger in your policy and system.

Yeah, so she's talking them about new ways that you don't even have to put that thing on there, you know, you can just come in the room and in the room has the environment, when the patient comes in, it just gives you like kind of the patient's whole thing, right, you can bring them in, and like 3d Walk around them, like they just kind of stand there, you can kind of walk around and like, Okay, this is going on here, your heart rates here, it looks like we got some distraction here on your phones. And so technology is far away. But, but it's getting there, right. And then this is where the future is going. And so our doctors like, you know, paying attention to it, you know, and likes her because she's like, interested in those things. My youngest daughter is more of the the one that I you know, I have more of impact on she is she's becoming a biomedical engineer and actually go to school, University of Florida. And she was about when she was about, I would say about 12. I had her writing, writing small scripts and Python, I used to do some, some robotics work out in University of Florida for Joel teen, she used to come visit me. And we'd sit there and and one of the cool things about it was that was Code Academy. For Python, my daughter's super smart. And the way Code Academy is set up. It's just it gives you a nice little introduction into programming, and Python syntax. And so you can start to slowly learn how to program and you can start to slowly understand Python syntax, which isn't so complicated. It's not like C++, you know, it's a little bit more, you know, syntax see heavy Python, you just like, you know, you make these senses, you make these things that is kind of make sense, right? If this happens, do this, that happens do that. And there's not a whole lot of syntax around it just kind of in which I learned that in school anyway. And she was able to, you know, to start developing, you know, smaller programs that could do our homework and stuff for you. And, yeah, and so now she's like, Yeah, Dad, I want to, I want to build robotic arms. And I want to, you know, build like these kind of like robotic prosthetics for people, right? Because she remembers when I was in the military, she and I got a buddy of mine who, who lost his arm in the military. And he's got a very, you know, it works. But it's really primitive, like a fork kind of fork finger arm thing. Yeah. And she wants to give them hands, right? She wants to, like, you know, hey, I want to be able to make and take the fork away and give him the ability to grab things. Yeah. And but, you know, what did she get that desire from the wanting to do that? And it was because of the accessibility of programming. And, and things like Python and ROS, write Python, being able to give you a really high level language that isn't overly complicated to start writing a program. It's very close to how you just kind of think and write things down when you're writing stories and things. And then ROS ROS gives you the ability to say okay, hey, I got a I got this module that does this, I got the camera, the camera takes in images. And I can get a process image. I've got the sensor that tells me what the height is up and down. And I can break that down into numbers. Essentially, you break all of those things down into numbers. And if you understand enough mathematics, you can now take camera images, you can take sensors, you can take lasers, you can take all these crazy things, you know that are you know, no one knows how, how that how the camera works, like how does it take the image? electronically? Yeah, exactly. That that's Oh my God, how long it will take you to understand, you know, yeah, it's like an undergraduate degree. Yeah, the sign What's behind that, but but if you can tell me hey, I can give you the you know, I can give you an array, you know, a vector of pixels and what those pixel values are, and those correlate to colors. And then you can decide to project that on to another screen that just like displays numbers, you can almost see the face that you're taking a picture of, even in ASCII, right, like we even without it even would actually give you the color values, I can give you an ASCII representation.

Yeah, or just use OpenCV, and then have it do face detection or some sort of thing, some other existing library, You learn OpenCV, and then you become your, you become a guru, right? Like you really get OpenCV. You gotta learn the math behind it, right? So you got to learn the math behind it. But you don't the tools, the tools give you the accessibility is the cool part. You make it accessible, you make it easy for these kids to get into it. And so and so you get the the little touches, I think I talked to you about this offline, was I had a kid, this this January, he was a was Cristo Rey has a high school program where they bring you know, other representative high school kids and merge them with a mentor from a tech company, to one talk this kid how to how to cut his camera on from a microphone, using Python, using Code Academy, quickly learned how to do let you know simple programs and modular programming, and bring in a library, this is the library for your camera. It's just how you cut your camera on. There's all these other API's in that library that can like focus the camera and go look at different things. But I'm just gonna, I'm just gonna show you how to cut it on and off. Yeah, and hit record for you. And then I'm going to show you this other library that takes your voice in and you can teach it words, but you know, teach it that camera teacher that on and it loads it from your voice, if I tell it ain't gonna work, but from your voice, it knows what cut means and has a camera on and knows what that means. And it sends the signal to the camera to turn on the camera API. And he's got that working in the kids. It's like, Oh, yes. Like, he was so upset about it, right? It's like, I love and see that. I love to see that because that kid's gonna grow, you know, I don't know he's gonna do in his life. I only mentored him for for about four or five months. But I showed him wonderful though, what is cake, what he's capable of doing? You know, and it was over four or five months. That was just, I just scratched the surface. And I always iterated. So this is just scratching the surface, if you actually want to be able to do this for real, and get paid, like I get paid to do it. You need to go to school, you need to learn you need to actually understand the math behind it, you need to understand the details behind this so that you do it the most efficient way the way we did it was a way they work, right. But this isn't the most efficient, the most effective way and ready for sure. And if you really want to get a job doing that, you can't go get a job making this right now because you're just a kid, right? But if you get a credential that says I know how to write software, well, I know how to develop cameras, and I know how to develop, you know, sound devices. And now you can go get a to this. Yeah, how cool is that? Right? How cool is it to go get paid to make these kinds of devices into make these things that affect the world. And he was totally, you know, he's totally stoked about it. And that that, to me is the key. That's the key to get our kids reback focused and, you know, technology and doing smart things. And I'm kind of bias. I'm a little older. I'm 43. And and I think like, you know, there's a good chunk of our kids. I think every generation has this, we just want to be like tick tock influences. And they want to you know, they want to do things that I like, Yeah, that's cool, when that's fun, but that isn't like, you know, progressing society. Right? Yeah, some way but you really need to be studying, you really need to be understanding, like standing on the shoulders of the guys that came before us, you know, with this technology and growing it. And it's not easy. You know, learning that calculus and learning that, you know, the physics... All the required skills.

That stuff is not easy. So how do you motivate those kids to stick with it when you know you got three dimensional calculus? You got integrals, and why am I doing this? Yeah, I could go vo here there was a tick tock dance. Making my money that way. It's like Nah, you get to actually once you learn that stuff, and what does that stuff apply to? That's another thing I think schools fail at is I agree completely explaining to you why is math important? Yeah, one plus one is soup. Okay, that's great. That's good for a simple math. Why is it that I need to know the integral of something what what usefulness is that? Yeah, this may give them something that is actually tangible.

Totally. This is why I think it's such an important thing that you're mentoring people. It's like It's hard to expect teachers to understand, like K through 12 teachers to understand the reasons that engineers use, say sinusoids, or linear algebra or whatever math related. Pythagorean theorem. Yeah, for sure.

All the programming concepts. But so like you who, as someone who actually does, it's very important for you to go hang out with kids, I think. And me too. Like, this is something I hope to be able to do in the near future, like, go help help help high school students or something like this. As someone who's actually applying these technologies, and be like, Oh, this is why sinusoids relevant, or whatever, which is something my teachers were not able to give me. When I was learning them. You learn them, you get the grade, you know? Yeah. I don't know how you got into engineering, but it wasn't because you're learning sinusoid like it wasn't your high school math teacher.

Not robotic for sure. I got into electrical because my dad told me I should. And then robotics was like, Oh, this is awesome. And so that's how I got into robotics.

Robotics is that's the cool thing about it is like, it's it's really world world, it's kind of hard sometimes for computer science, you know, to kind of explain, okay, yeah, you make this website. But in order to make this website, you got to know all these, like data structures, and all this other stuff that is not necessarily website base. And then in the, you know, the backend stuff is a little bit more complicated, but you know, being able to tie it to something that's real world, and I think robotics, is a very cool thing, because there's just so much, especially right now, there's so much room for things that are not there. You know, when when kids are usually the engineers, or the kids who are like, I wish this could work like this, we got this thing already. But why couldn't they work like this? Like who, you know, I'm not the guy who made it, but why did they make it like that I would have made it this way, I would have been something like this, I don't know how to do it. But if I could, I would have made it this way, I would have made my camera, you know, fly around on a drone and follow me around the house, right? Like, and then I don't have to just sit in my chair, and have this video, I could just go for a walk. And we could have this video, right? That'd be super cool. You know, and we got it, we have that technology now. Alright, but if no one ever developed that and thought of that up, that wouldn't have happened. And, and we got to give those kids the, you know, I think early on the early on is a photogram. Because the technology is accessible. Now, it's, it's not as complicated. It's not like our grandfathers and our fathers, you know, entry level to get in. It is now at a level where I can teach my seven year old daughter, you know, hey, this is how you program a little math equation in Python. Yes, how you take variables. And this is how you make these variables do this map. And this is how you get a result out.

2 posts with the testing tag