Charlie Andersen - Burro - Ep 29
In this episode of "Let's Talk Farm to Fork", we're joined by Charlie Andersen, from Burro, who we will be talking to about how their autonomous robots are helping optimise field labour for fresh supplies through carrying, towing, scouting, and beyond.
Transcript
[00:00:00] Mitchell Denton: Hi there, and welcome to "Let's Talk Farm to Fork", the PostHarvest podcast that interviews people of interest across the food supply chain.
Today on our show, I'm joined by Charlie Andersen, from Burro. Who I'll be talking to about how their autonomous robots are helping optimise field labour for fresh supplies through carrying, towing, scouting and beyond.
So with no further delays let's get started.
Well, hello, Charlie. How are you? Thanks for joining me on the podcast today.
[00:00:28] Charlie Andersen: Yeah Mitch, great. Great to be here. Really appreciate the interest and, and, uh, great to be here.
-
[00:00:32] Mitchell Denton: Of course, of course. Before we get into it, I just wanted to give you the opportunity to tell us a little bit about yourself, and what you do, and maybe just a fun fact about yourself.
[00:00:41] Charlie Andersen: Yeah, sure thing. So, uh, so my name is, Charlie Anderson, um, I grew up on a, I guess, a working fruit and vegetable farm in Pennsylvania on the east coast of the US. And my, I guess my, my kind of lifelong obsession has been trying to figure out ways within farming of doing work from a tractor cab and not having to get out of the tractor cab to go to work by hand.
Um, and that interest has led me to where I am today. Today I run a company called Burro. Uh, we build, you can effectively think of it as, as Disney's "Wall-E for agriculture" in a 1.0 format. So a computer vision-based, autonomous ground vehicle. That does a variety of tasks outdoors and is designed to lay the base for a lot more autonomy over time.
Um, and in terms of fun fact, I guess, as I thinking through things, I, I, on my family farm, I built a road that is now mapped on Google Earth. So that's one, one fun fact, if you will.
[00:01:38] Mitchell Denton: Which road is that? If you don't mind me asking.
[00:01:40] Charlie Andersen: It's a, um, it's a gravel road that goes about like three quarters of a mile up a hill. Uh, and, um, did it with a, with a, basically a small bulldozer.
[00:01:52] Mitchell Denton: Yeah. Wow.
[00:01:53] Charlie Andersen: So, uh, so not yet, not one that you could just, you know, randomly get on, get into a car and go on, but, but still sits on Google Earth, which is somewhat of a bragging right for me, if you will.
[00:02:01] Mitchell Denton: Yeah, that's pretty cool. All right, well, before we get bogged down in road talk, let's talk fun to fork. So, continuing on from you telling us what you do. I really enjoyed that Wall-E definition that you gave of your technology, would you mind telling us a little bit more about the history behind Burro and how your innovative technology works?
[00:02:22] Charlie Andersen: Yeah, sure thing. So again, grew up on a working farm. I got an MBA and got out of business school, went to go work for CNH, which is Deere's largest competitor. And there part of my role was selling and marketing machinery to farmers.
And the other portion, uh, I spent a fair amount of time looking at autonomy companies from an acquisition perspective for the company.
And from like a really, really kind of frustrated with how slowly robotics was making its way into agriculture and kind of how almost how uninterest large incumbents like some of the ones I was working for, um, were, were moving.
And so I had a, I believe I had a colleague whose family had a chicken farm, and idea number one became, "Hey, I'm gonna go figure out how to build a robot to pick up dead chickens. And, and needless to say that we, we've pivoted from that concept...
[00:03:10] Mitchell Denton: Yeah.
[00:03:11] Charlie Andersen: Quite a bit. But as I was looking at that idea, what I was discovering was that there are all sorts of applications within agriculture. Where you have people moving through the world, perceiving stuff around them, and then manipulating things.
And if you can do the mobility part and possibly the perception part that platform can get into a ton of different use cases. And, and that kind of led me to, again, what our product is today, which is an autonomous ground vehicle used in primarily in vineyards, nurseries and berry operations, but get kinda getting pulled into a bunch of other sectors.
[00:03:48] Mitchell Denton: Hmm. Yeah, no, that's great. So what would you say separates Burro technology from other labour solutions currently on the market?
[00:03:56] Charlie Andersen: Yeah, totally. So, there's this thing within robotics called "Moravec's paradox", it's, it's the notion that it's a lot easier to make a computer that can play chess, than it is to make a computer or a robot drive across the room. And so said another way, if you're a robot on a farm, you have to one understand where are you in the world?
You have to then understand how to move from A to B, what's around you, and then ultimately, how do you manipulate those things around you? And I think the way a lot of companies have tried to tackle autonomy on farm to date has largely been by automating a tractor or by trying to go out and pick fruit.
And if you look at other segments such as warehousing or factories, the way robots have initially started has been by picking a very constrained type of mobility and beginning with that mobility. But later over time doing more things. And so, what we have done is we've built what's basically a four or five horsepower autonomous farm ATV.
That can carry around 400 to 500 pounds around 225 or 230 kilograms. And what that vehicle does is it can be operated by any person on a farm. It can run without any sort of central control system.
And it uses a combination of computer vision and high persistent GPS, plus a ton of, uh, artificial intelligence to follow people and structure within an environment to learn roots and effectively to carry or tow heavy things from A to B to C and to perceive things around it or to scout things around it.
And so you, you might ask, so what's the use case for a product that moves from A to B to C alongside people within agriculture? I think what, what we have found is that, within ag, there are all sorts of settings where you've got people working under canopies, so you can't just use GPS to navigate. And you've got people that are carrying heavy things around and the people that do a lot of labour on farm, tend not to be equipment operators.
They tend to be, you know, relatively, um, uh, less sophisticated from a machinery perspective. And so what separates our system from pretty much everything else on the market. Is that its operating under canopies and out in the open. So with GPS and without GPS and it's working safely alongside people to carry heavy things.
And that's, so we're, kind of a weird duck in the marketplace where not many other companies have something like ours and we actually can't quite figure out why that is the case, we're just a weird company in a way.
[00:06:36] Mitchell Denton: Yeah, no, that's cool. I, I, I really love the marriage of complexity and simplicity with the technology it's such a, um, a straightforward application, obviously there's a, there's a whole lot of backend work that goes into that simplicity, but it's just a really cool, uh, solution at the end of the day, it's really cool.
So, Agriculture and AgTech seems to be a market that is currently garnering the attention of big VC firms these days. In fact, I see that Toyota Ventures has recently contributed to funding to Burro.
Are VC's attracted to AgTech because of its frontier technology aspect or more because of its sustainable outcomes?
[00:07:17] Charlie Andersen: Yeah. So I'm trying to think of how to, so how to answer that. So there, there are pros and cons of agriculture. So when you think about, when you think about Ag, specifically, half of revenue tends to be crops and half of revenue tends to be livestock.
Within crops, you've got grains, so corn, wheat, soybeans, and those at least within the US are something like 70% of farm revenue. But only about 8 to 10% of farm labour. And then you've got all the specialty crops and those are roughly 30% of farm revenue, but 80 to 90% of farm labour. And that, that disparity has for a long period of time, really scared people away from investing in Ag because where all the labour is, tends to be in these kind of nichey, funky different use cases that are really difficult to automate comprehensively.
And if you automate one use case comprehensively, ie pick strawberries, then you're only able to pick strawberries and only in one production style and can't move to nursery crops or, or cherry tomatoes or something else. Um, so that's kind of the, that's the kind of why have a lot of investors shied away from it historically from a TAM perspective?
The other reason that people have shied away from it has been that it's really, really difficult to do. So I think it's, it's remarkably easy to prototype something that works in a demo or kind of a video type context. But as you go from 1 to 50, to 100 to 200 systems running your system has to handle a ton of additional variability.
And that variability increases almost exponentially for each additional unit you have running, which means it's very difficult to build something that's reliable, and can do what people actually expect. So those are the kind of, why not thus far, small nichey, funky TAMS, and then really, really difficult to get the technology working.
The flip side of it is that the difficulty also exists on road or in a lot of these other segments and Ag, you can move slowly, you're off road, you tend to be away from tons and tons of people, or at least the public. And there's a huge labour dynamic that is begging to be tackled.
And the second thing, so, so not only is there a really pervasive, specific need. I think the second thing is that the need can be met increasingly by computer vision based autonomy. And so from all of this, nobody has succeeded yet. There's all of this potential demand that people can kind of see. And, and the technology is suddenly there.
And so I think some of the companies that operate today, ours hopefully including, will be really, really significant companies ultimately, but it kind of has a feeling of like, "Mount Everest in 1948" or like "PCs in 1976", you know, it's early stage, early days. Nobody has succeeded yet, but you can kind of sense that there's something on the horizon.
I think that, that's what Toyota and F-Prime and Cibus and, and some of our investors amongst many others are starting to really, to really see.
[00:10:20] Mitchell Denton: Mm. Yeah, so then what would you say is the biggest challenge your team has encountered so far with your innovative robots and how did you overcome it?
[00:10:29] Charlie Andersen: Um, localisation is the hardest problem that we've encountered bar none. And it's, it's something that's obvious, it's everywhere, and it's so obvious that people don't even see it. But it's how do you answer the question of where are you in the world when you're going out in the open and under canopies, and that if you can't reliably answer, where are you in the world?
It's very difficult to build a system that can behave really reliably. Um, and just to, to unpack that a bit more, you, you have, even fairly simple things that seem simple to your reptilian brain, uh, that are incredibly difficult to do for a robot. If you're living from A, to B to C in an outdoor setting, and you go from out in the open to under a canopy, you go from having perfect GPS to no GPS or GPS that is bouncing around like a, marble in a blender with the top off.
So you're, you've got one signal in your system that's bouncing around all over the place, and then you're going from bright light to dark light. And you tend to be doing it in environments where tons of stuff is moving around. Uh, so lighting is changing all day long, shadows are moving around, things are moving around, et cetera.
And so I think for us, the obvious thing is that localisation is really, really difficult to do well. And from that, our assumption is that if we can do it really, really well, then our system can become a platform for others to use. And separately from that, it's a land of no silver bullets, only lead ones.
There's no like, one single solution that solves at all. It's, it's you have to run 30, 50, 100 thousand plus miles or, or, you know, a million plus miles with systems and real world conditions and encounter edge cases constantly, and just kind of gradually hammer away with different solutions to all the kind nichey specific problems that come up.
So yeah, so that's a very long wind explanation, but localisation is really hard, difficult to do well, and in terms of how we have accomplished it or overcome it thus far, we have not overcome it in every single use case thus far yet.
Our systems function between 15 and 20 autonomous miles per user intervention today.
And at those types of rates, roughly 30% of our faults relate to robots being lost. When they're lost, they stop. So the problem is largely mitigated. And what we discover is that as we go into new environments, the way you solve the localisation problem changes somewhat. Although the same stack can ultimately build work across everything.
[00:13:01] Mitchell Denton: Yeah wow. I mean, continuing on this thread of biggest challenges. While working in AgTech, what have you found to be the biggest surprise?
[00:13:09] Charlie Andersen: You know, to me, it's just all about the people. I think it's kind of the, the, the, this is ultimately, technology and technology within AgTech is a pure people business. You need an incredible team to build a product. That team needs to be really, intrinsically motivated because no single person has all the answers of how to solve those problems.
And then you need to build product that end users love and can figure out how to operate. And I, you know, as somebody who should be considered classically trained in this domain, I grew up on a working farm. I, I worked for a big Ag equipment company.
It's definitely been a surprise to me, how much of a people business I'm in, uh, and, and largely when we have, we have problems or opportunities, most of them centre around finding great people, getting them really, really motivated and really satisfying an end user with what the product does.
[00:14:03] Mitchell Denton: Yeah, absolutely. I, I would take it a step further and say even beyond like the inner team, just the groups of people within the industry that are more than happy to collaborate and really want to see success just across the board.
[00:14:16] Charlie Andersen: Totally, and it's, I think it's, it's also surprisingly small. Like you run into these events. You have somebody that that's familiar because you've seen them on a website somewhere. Like it, it, it, I think it very, very quickly builds. Like I, four years ago, I'd fly out to California, like sleep in the trunk of a KIA and go meet with people.
And a lot of the people I, I met from four years ago have just kind of like spiraled into a million other connections. And typically like the second or third person I meet in the space leads me back to somebody I met previously. I think it's a really, it's just a, a really cool industry from that perspective with tons enthusiastic, bright, driven people that, that really love what, they do.
[00:14:49] Mitchell Denton: Yeah, totally. Totally. So, from your perspective, what do you identify as being one of the biggest pain points or blind spots in the food industry?
[00:14:59] Charlie Andersen: I'm not sure if it's a, if it's a blind spot more just of a pain point, I think that within our space, people talk about labour being a pain point.
And I think that, that labour is a real pain point, but it has a bunch of layers to it. I think that the real pain point is that nobody wants to get up anymore and go into a field at 5:00 AM in the morning and work from dawn to dusk in 110 degree heat.
And the people that are still willing to do that work increasingly, um, are demanding better conditions and also are wanting to be kind of upskilled to be able to do more per unit of time than they kind of do.
So again, I think, I think labour is a big pain point. I think it's described typically kind of in an abstract way and you really need to think about like, "Who's doing the work?"
And like, like I, I, I will encounter crews all the time where like, you know, there'll be a father, mother, son or daughter team. And like the parents are working incredibly hard cause they want their kids to be better.
And then they'll pull their kids in the field to show their kids how hard field work is to inspire their kids, to do something else or to work harder in school or do something. And so labour is a big pain point and the blind spot in all that is there are people working on these farms, you really gotta go talk with them to understand, basically to understand their dynamics.
[00:16:15] Mitchell Denton: Absolutely. So, when it comes to food loss and sustainable farming, what's the biggest area your team are curious about and why?
[00:16:25] Charlie Andersen: The English economist, Thomas Malthus, he talks about how at some point, all, all populations reach the point at which they're consuming all resources or they're over consuming resources in the environment can no longer sustain them.
And I think that in the case of people, if you fast forward the clock on where human population is going, in the next couple of years, or say the next 30 years or so, we are supposed to reach the point where we are consuming three X, what the planet produced in 2010, which, which obviously is not going to happen.
And so to me, what I'm most curious about, I think, I think in the past we've had the mechanical revolution, we've had a chemistry revolution, we've had a, a genetically modified organisms revolution and all those things. And to me, what I'm most curious about is how big of a picture can automation play in basically doing three X more in food production than what we're doing today.
And, and if you, if you think about like, what do you do in your, what do you do in your garden? When you see a beetle eating the leaf of a tomato plant, well you squish it and you do that on a plant by plant basis, you're, you're manipulating an individual thing.
You get into big production Ag, you can't do that, you've gotta do the same thing on everything. I think that being able to see something and behave very specifically to it in real time has this huge potential to alter the way we actually produce food. And I'm very curious about how that takes place.
And then the final question and all that is, "is all that, all that done by one company?"
And I don't think it is. I think it's actually done by many, many different companies and some will be building mobility, others are gonna be building things around the neural networks to identify particular diseases and there'd be ones working on manipulation and ones doing distribution.
Um, so yeah, there's a lengthy list of questions to what was probably intended as a 30 second answer, so those are something of the things that are trickling through my mind.
[00:18:17] Mitchell Denton: No, this is the good stuff. This is why I asked the questions. So continue on with this thought. Is there a particular group or innovation within the industry that you're excitedly keeping a watchful eye on?
[00:18:29] Charlie Andersen: So, going back to the layers comment, I think there's a, there's a group called Western Growers Association. And they have a perspective on autonomy that, that things are ultimately gonna be built in layers.
Um and, I really share that perspective. I think that that will also take place, um, at the same time, with the way the industry is today, a lot of the groups are very reticent to work with others, because it's unclear who's a competitor and who's a, who's a, who's a friend, who's a friend who's a fo is kind of unclear.
Um, and so what I'm particularly interested in is following the work that Western Growers Association is doing to encourage layers and then trying to figure out what we can do as a company to build the ultimate mobility platform on top of which other companies might build their manipulation system or crop perception system, or... I had a guy email me today with a, with a system that's supposed to navigate around and feed animals.
There are all of these funky ideas where I don't think that one company does them all. I think they're actually layers. And I'm kind of curious how the industry and how companies in the industry can advance that.
[00:19:35] Mitchell Denton: Yeah. Yeah. So then what's one thing you wish you'd known when you began your career in developing autonomous farming robots?
[00:19:44] Charlie Andersen: Um, I think it's just really hard and that's that's, um, it's, it's hard. People talk about barriers to entry and, and patent protection and, and things like that. And, and the realities of building these systems is so incredibly hard. I think that had I known that from the outset, I probably wouldn't have done what I'm doing today and it probably would be... like, I I'm glad that I'm doing today what I'm doing today, but that, but that said, I think it's really, really hard. And I think just kind of recognition of what's the stuff that is, it's not obvious from the outset, what is going to be hard as you get into it.
[00:20:23] Mitchell Denton: Yeah. Absolutely. Absolutely. So Charlie, we are unfortunately coming to a close, but as this episode draws to a close, is the main point you want the listeners to take away?
[00:20:35] Charlie Andersen: Yeah, so I think, people are kind of oddly good at predicting where the world is going in 10 to 20 years, but really, really struggle to predict how do you win in years 1 through 5. And to me, the main point, at least as it, pertains to my company in this topic. Is that I think that in 10 to 20 years, you're gonna see millions of robots doing a variety of tasks on farm and elsewhere.
I think that the approach that we are taking, beginning with a mobility platform, starting in vineyards and nurseries and berries is a unique approach that appears to be proving successful today. And that if companies are seeking to build layers around that require mobility, require perception or, or require moving manipulation from A to B, we would love to partner with them.
And then I think that again, I think fast forward two decades from today, and again, the same way that you have a Roomba navigating through your kitchen, vacuum cleaning. I think you're gonna have tons of small autonomous systems in the world's largest industry doing a whole host of different tasks.
And so it's almost like we're sitting in, we're sitting in 1977 or 78 as early, early main trains and PCs are emerging and we're trying to envision Twitter and that's a really, really exciting place to be. But also we're very much kind of predicting something before it's happens, which again, exciting.
Paving the way for a future, solving huge problems, but also a, uh, an unpredictable place to be in the world.
[00:22:12] Mitchell Denton: Absolutely. I look forward to seeing what lays ahead for the next few years.
[00:22:16] Charlie Andersen: Yeah, me too.
[00:22:17] Mitchell Denton: Well, that's all for today's episode "Let's Talk Farm to Fork". Thanks for listening. And thank you, Charlie, for joining me today.
[00:22:23] Charlie Andersen: Great, Mitch. Thank you so much, we really appreciate being on. So thank you.
[00:22:27] Mitchell Denton: If you'd like to know more about Charlie and Burro, check out the link and the description on the episode, make sure to subscribe to the podcast so that you never miss an episode.
And don't forget to leave a review and share with your friends.
Until next time you've been listening to "Let's Talk Farm to Fork", a PostHarvest podcast.
Other Episodes
