Latest / Automated with Brian Heater / Physical AI, Play, and the Path Beyond Humanoids with Brad Porter of Cobot

Transcript

Physical AI, Play, and the Path Beyond Humanoids with Brad Porter of Cobot · Automated with Brian Heater

• 0:00
  SPEAKER_01: You think about just like washing dishes, right?
• 0:03
  SPEAKER_01: And sure, we can kind of rot train things, but understanding the texture of a dry sponge versus a wet sponge, understanding how to flip open the lid of a dish detergent because each one's a little bit different, like it is physically problem solving in the real world.
• 0:21
  SPEAKER_01: I think we need to imbue robotics with that ability to kind of learn things through trial and error.
• 0:27
  SPEAKER_01: As we unlock that, we won't be playing around in simulation.
• 0:32
  SPEAKER_01: The robots will just be trying new things in the world like you or I do.
• 0:56
  SPEAKER_00: Prior to Cobot, Brad spent 13 years at Amazon, where he was the VP of robotics and helped lead the rollout of the world's largest fleet of industrial mobile robots.
• 1:10
  SPEAKER_00: Brad was also recently a fireside speaker at our humanoid robot forum in Seattle, where he provided a very entertaining counterpoint to the ongoing humanoid hype cycle while discussing the company's proxy robot.
• 1:23
  SPEAKER_00: It's always a good time speaking with Brad, and this podcast uh certainly isn't an exception.
• 1:30
  SPEAKER_00: Don't forget to like, subscribe, and tell a friend, all of that good stuff.
• 1:35
  SPEAKER_00: Enjoy.
• 1:36
  SPEAKER_00: So, as we were programming our humanoid summit agenda, somebody suggested that we have a panel or presentation that was kind of a reality check as far as the subject of humanoids.
• 1:50
  SPEAKER_00: Two people immediately came to mind.
• 1:52
  SPEAKER_00: The first one was recent podcast guest Broad Brooks, and the other one is sitting across from me right now.
• 1:58
  SPEAKER_00: How does that feel that you're like the one or two person that people think of when they think about skepticism around this form factor?
• 2:07
  SPEAKER_01: You don't like go out to be a skeptic.
• 2:09
  SPEAKER_01: Like there's kind of a lot of enthusiasm and excitement about humanoids.
• 2:12
  SPEAKER_01: I felt like it was worth putting out some content to explain what I had seen, you know, thinking deeply about humanoids at Amazon.
• 2:20
  SPEAKER_01: I think, you know, a little bit ahead of the curve, and then I thought about this a lot in 2018.
• 2:25
  SPEAKER_01: I think agility had showed some really impressive bipedalism.
• 2:29
  SPEAKER_01: That technology looked like it was starting to be viable.
• 2:32
  SPEAKER_01: The the humanoid grand challenge had just taken place.
• 2:35
  SPEAKER_01: Um, and so we thought about it a lot.
• 2:37
  SPEAKER_01: And then we came to the other side of that exercise realizing that for the use cases where a humanoid would be would be helpful at Amazon, it wasn't actually the best form factor.
• 2:48
  SPEAKER_01: And so that led me to realize there was really what we needed in commercial environments was quite a bit different.
• 2:55
  SPEAKER_01: And then in the home space, I never quite figured out exactly how humanoid helps us either.
• 3:01
  SPEAKER_01: Um, and so then you know, I put out content on this, and then you know, of a sudden I'm I'm like branded the kind of skeptic because there you, you know, on any on any interesting story, people are always looking for for two sides and to create this kind of polarity.
• 3:15
  SPEAKER_01: And um, and so yes, I guess I've I've ended up anchoring the uh the maybe more pragmatic, uh, what what do commercial environments really need?
• 3:25
  SPEAKER_01: Um, and and pouring a bit of a dose of cold water on the humanoid hype.
• 3:31
  SPEAKER_00: Well, when you're saying that for any given thing that the robot would be doing um, like in in a warehouse specifically, they're maybe not the best suited for that.
• 3:41
  SPEAKER_00: Are you are you talking specifically about a robot with two legs when you talk about humanoids?
• 3:46
  SPEAKER_01: As we looked at humanoids, and as I continue to look at uh robotics, there were a few things that that stood out as as really critical in commercial environments.
• 3:57
  SPEAKER_01: One is you want to be safety-rated and safety-real, because most of these process paths that would be that aren't going to be automated in some other way benefit from being able to work in and around humans.
• 4:09
  SPEAKER_01: And if you're trying to work in and around humans, you don't want any risk of falling on a human, injuring a human.
• 4:16
  SPEAKER_01: Um, and when you have a balancing system, that's very tough to guarantee, right?
• 4:24
  SPEAKER_01: Because any glitch between sensing and electrical and actuation can cause an instability that causes that robot to fall, right?
• 4:33
  SPEAKER_01: And so that was the first thing was, and and then you ask yourself, do you really need bipedalism?
• 4:39
  SPEAKER_01: And then, you know, in in commercial environments, people are moving things around with the assist of wheels, of carts.
• 4:45
  SPEAKER_01: So you're not carrying things up and down stairs that you know that might exist in apartment buildings in New York City trying to move in and out, but that doesn't exist in in a commercial warehouse, right?
• 4:56
  SPEAKER_01: Nobody's really carrying things up and down stairs.
• 4:58
  SPEAKER_01: There's not a real need for bipedalism.
• 5:00
  SPEAKER_01: Yes, there are ramps and lips and things like that, but you can you can get over those with with wheels, with with maybe some degree of suspension, active or passive suspension.
• 5:09
  SPEAKER_01: So you you don't need bipedalism, you don't need quadrupeds either, right?
• 5:14
  SPEAKER_01: Like good wheels, the right size wheels, the right size suspension will will help you in in whatever you need to do.
• 5:21
  SPEAKER_01: You really want swappable batteries?
• 5:23
  SPEAKER_01: Nobody in humanoids is really solving.
• 5:25
  SPEAKER_01: Like if you have to sit on a charger, you lose, you know.
• 5:29
  SPEAKER_01: Look, we as humans, we sleep seven, eight hours a day, we lose a third of our productivity every day to sleep, right?
• 5:36
  SPEAKER_01: Why would you do that in an industrial environment if you could avoid it?
• 5:39
  SPEAKER_00: It's why there lose productivity or you have to buy like three times as many robots.
• 5:43
  SPEAKER_01: And you need a whole bunch of space for the robots to park as they charge.
• 5:47
  SPEAKER_01: Like, really, like I'm even more bullish that like sitting on a charger is uh, or maybe more opinionated that sitting on a charger is a completely horrible use of a robot.
• 5:57
  SPEAKER_01: Your whole goal of a robot is to be operating in production continuously and taking on meaningful work, and so sitting on a charger doesn't make any sense.
• 6:08
  SPEAKER_01: Um you really want your sensors not to move around and not to move up and down.
• 6:13
  SPEAKER_01: So the problem is if you're a balancing system, your sensing is always moving around, and so you have to you have to then compensate with your AI algorithms for the fact that your sensors are never stable and so your calibration's a lot harder.
• 6:26
  SPEAKER_01: Yeah, they're just a whole bunch of reasons, and the cost, the amount of, and and then I think the the biggest problem is they're not strong enough, right?
• 6:35
  SPEAKER_01: The you can't get the torque density in one of these like shoulder motors to be able to lift what humans can lift.
• 6:42
  SPEAKER_01: And look, we use linear actuation in our muscles, and look, we're very, very finely tuned for it.
• 6:47
  SPEAKER_01: You know, our robot proxy, which is a you know, is more of the form factor that obviously I espouse four wheels on the ground.
• 6:54
  SPEAKER_01: The latest generation can lift 200 pounds, right?
• 6:57
  SPEAKER_01: Now, it does that because it has a screwdrive in the spine, right?
• 7:02
  SPEAKER_01: Which is a much better way to get torque, right?
• 7:05
  SPEAKER_01: Now, if you or I had a screwdrive in our spine, like we could lift a lot too, right?
• 7:10
  SPEAKER_01: And so you actually want robots that are more capable than humans, right?
• 7:15
  SPEAKER_01: You want to move 1500-pound carts around that helps to have like four electric wheelchair motors in the base, right?
• 7:25
  SPEAKER_01: That can move 1500 pounds without any trouble.
• 7:28
  SPEAKER_01: You or I moving 1500 pounds, we can do it, but we're using a lot of shoulder, a lot of quads to try to get that moving.
• 7:34
  SPEAKER_01: Um, and I the humanoids just don't quite have that strength.
• 7:37
  SPEAKER_01: And so where I became kind of disillusioned was you didn't need bipedalism, you needed safety rating, you wanted battery swap, you wanted something really, really strong, you wanted the sensing to be stable, and you were just never going to get that if you constrained yourself to the human film factor.
• 7:55
  SPEAKER_00: Yeah, that this I'm I'm glad that the people are really having the safety conversation right now because I think maybe something a lot of people, or at least a lot of people don't uh outside of the industry don't realize is that like I think to this day, the vast majority of the AMRs at Amazon are still in a caged environment.
• 8:12
  SPEAKER_01: The Proteus is not in a caged environment.
• 8:15
  SPEAKER_01: It's a it's a beautiful piece of engineering.
• 8:17
  SPEAKER_00: That's the autonomous version.
• 8:18
  SPEAKER_01: Yeah, it's their green autonomous one with the kind of blinky eyes that chirps at you.
• 8:22
  SPEAKER_01: It's that that that program was uh actually pre-jated me a little bit when I was at Amazon, and then was something we invested in heavily at Amazon and and they they got into production after I was after I had left.
• 8:36
  SPEAKER_01: But uh super impressive robot.
• 8:38
  SPEAKER_01: I I I I love that robot, but it has to perceive the world from the ground, right?
• 8:44
  SPEAKER_01: And trying to perceive, I mean, imagine you or I try and understand what's going on in a warehouse around us if we had to lay on the floor and look up from six inches.
• 8:53
  SPEAKER_01: Like it's very hard.
• 8:54
  SPEAKER_00: So I'm guessing they would build warehouses very differently if we were that close to the ground.
• 8:59
  SPEAKER_01: Yeah, I think so.
• 9:01
  SPEAKER_01: I think so.
• 9:02
  SPEAKER_00: It's interesting to hear you say that at that early stage at Amazon that you were having this conversation around humanoids.
• 9:07
  SPEAKER_00: Like obviously, right now, as things are bubbling up, there's all these reports coming out that you know Amazon is more actively looking at humanoids.
• 9:14
  SPEAKER_00: We know that they had the pilot with agility a couple of years ago.
• 9:18
  SPEAKER_00: What did those what shape did those conversations take that far back when discussing humanoids?
• 9:26
  SPEAKER_01: Look, Amazon, when I was there and continues like uh is in a position and has a almost imperative to to understand all of the latest technology, right?
• 9:40
  SPEAKER_01: And how it can be used to optimize Amazon's operations, right?
• 9:44
  SPEAKER_01: The you know, from a shareholder standpoint, that's what shareholders would expect.
• 9:50
  SPEAKER_01: From a customer standpoint, we continue to want our stuff faster, lower cost, right?
• 9:55
  SPEAKER_00: And so But it's like it's like a balancing act, right?
• 9:57
  SPEAKER_00: It's like being as forward-thinking as possible, but not actually implementing any of that technology until it's ready to work full-time.
• 10:05
  SPEAKER_01: Well, I think that that's always the the question, right?
• 10:08
  SPEAKER_01: Always the question is is this ready now?
• 10:12
  SPEAKER_01: Is it something we can help accelerate?
• 10:14
  SPEAKER_01: Um, is it something we can help accelerate?
• 10:18
  SPEAKER_01: By we, I'm I'm speaking as as if I'm still at Amazon.
• 10:21
  SPEAKER_01: I left Amazon 2020.
• 10:23
  SPEAKER_01: But yeah, you know, the question was could we help accelerate it?
• 10:28
  SPEAKER_01: Should we develop some of that cable?
• 10:30
  SPEAKER_01: So for dr drones, for example, Amazon decided that the best way to accelerate that was to invest in in building them themselves, right?
• 10:38
  SPEAKER_01: In the humanoid space, Amazon's you know made big investments, right?
• 10:42
  SPEAKER_01: We didn't have a humanoid in-house investment.
• 10:45
  SPEAKER_01: Whether they do now or not is uh there's speculation, but there's there's no confirmation that I know.
• 10:52
  SPEAKER_01: Um, and so you know, there they decided to invest in the partner ecosystem.
• 10:56
  SPEAKER_01: So look at Amazon, like our view was we just needed more robotic capability, right?
• 11:03
  SPEAKER_01: The more capability out there, the better.
• 11:06
  SPEAKER_01: And then we as Amazon could choose which technology's partners we felt had the most potential and could go deeply.
• 11:15
  SPEAKER_01: And some of them were more speculative and some of them were less.
• 11:18
  SPEAKER_01: And so we really liked the uh agility team, and so I'm not surprised that that relationship continued after I left, and I think it expanded after I left, and as far as I know, is it's continuing.
• 11:31
  SPEAKER_01: Um, but we also, as I said, knew it was speculative.
• 11:34
  SPEAKER_01: I think from my perspective, I started to become disillusioned that it was the right solution for a lot of these use cases that Amazon had, but that hospitals have, that manufacturing has, these use cases that are really, you know, human-scale material movement, moving boxes, totes, and carts around the world, I thought there was a much better form factor.
• 11:57
  SPEAKER_00: I I don't know if you're having these conversations right now, but at the points in uh Cobalt's progress when you were having conversations with VCs, um you know, especially the last couple of years, has there been push for you to make proxy more humanoid?
• 12:15
  SPEAKER_00: Because that's like the hot thing that everybody's seeing and and and and shoving money into right now?
• 12:21
  SPEAKER_01: Does the presence of like multiple solutions it does two things, right?
• 12:25
  SPEAKER_01: One, it both drives interest in the space, which is exciting.
• 12:30
  SPEAKER_01: Right?
• 12:30
  SPEAKER_01: I mean, we we raised our series B about the same time figure raised its big big round, right?
• 12:36
  SPEAKER_01: And that was a little bit just because there was a lot of investor attention in that space.
• 12:40
  SPEAKER_00: Rising tide, yeah.
• 12:42
  SPEAKER_01: Investors can only have so much attention and bandwidth, right?
• 12:46
  SPEAKER_01: And so they can only be focusing on so many segments at once, and so that created a little bit of a crucible where everyone was focusing on on robotics for a bit.
• 12:55
  SPEAKER_01: Now, the interesting thing is what some of the top investors in the world will tell you uh I'm not gonna name them because then maybe I'm putting words in their mouth, but what's some of the top is that you to make a really successful investment, you need to invest in something that's contrarian and right.
• 13:14
  SPEAKER_01: Right?
• 13:15
  SPEAKER_01: So think Airbnb, think Uber, right?
• 13:18
  SPEAKER_01: These things were contrarian, they not everyone believed in them, not everyone bought into the idea that we were gonna sleep on each other's couches, right?
• 13:25
  SPEAKER_01: Um, and yet they were right.
• 13:27
  SPEAKER_01: And so that's a little bit the argument behind behind Cobalt and Proxy, is like, okay, Brad's a little bit the contrarian here, but is he right?
• 13:38
  SPEAKER_01: Now the problem is you have multiple perspectives out there, you have these things that are really exciting.
• 13:44
  SPEAKER_01: Um, and so that also creates like um some cognitive dissonance, right?
• 13:49
  SPEAKER_01: And some level of like confused minds don't always don't always invest either, right?
• 13:54
  SPEAKER_01: And so, you know, I think the the landscape is still shaping out.
• 13:59
  SPEAKER_01: What I would say is in the past year, it certainly feels like you know, what we've seen is a proliferation of humanoid companies.
• 14:08
  SPEAKER_01: So the idea that what some of the folks maybe like figure did where they put together a humanoid very, very quickly, seemed like shockingly impressive.
• 14:17
  SPEAKER_01: Now we see like teams of five in a garage coming up with a humanoid.
• 14:21
  SPEAKER_01: It it's it's not it was an impressive technical feat, it's impressive when the teams of five do it too.
• 14:28
  SPEAKER_01: Um, but you start to realize, well, wait, it it wasn't maybe that hard.
• 14:34
  SPEAKER_01: The other thing that happens when investors get interested in a space and then they're they're maybe confused about it for a bit, is they study it really, really hard, right?
• 14:43
  SPEAKER_01: And they kind of learn the whole space and they consume everything, they read everything they possibly can, they talk to everyone they possibly can, and then they get a much more refined thesis um as to what might work and what might not.
• 14:55
  SPEAKER_01: And I think you you see now that there's a little more caution uh in thinking about humanoid investments now, and but I think also, you know, a little bit of a dose back to reality of like how hard robotics is, right?
• 15:11
  SPEAKER_01: And I think you know, investors have always been really excited about robotics, but robotics always goes through these kind of like ups and downs where people realize oof, it's really, really hard, right?
• 15:21
  SPEAKER_01: And I think it takes a fair bit of capital intensity and the deployment cycles are a little bit longer, and so I think we're in a very good position.
• 15:29
  SPEAKER_01: Um, but it it is it is tough in robotics right now for a lot of folks.
• 15:35
  SPEAKER_00: Is there a sense in which proxy is the more cautious bet versus a lot of these legged robots?
• 15:42
  SPEAKER_01: Well, I think in some ways it's the more ambitious bet, right?
• 15:45
  SPEAKER_01: Like it's it's easier to kind of follow the trend line, say, oh, you know what, I can build that, but I can build it with a better shoulder actuator or something like that, right?
• 15:59
  SPEAKER_01: I think what we've done in trying to bring a uh a conceptual model that people haven't seen before.
• 16:07
  SPEAKER_01: There's a kind of product adage that you should, if you're bringing out something new, make it look a lot like something that already exists, right?
• 16:15
  SPEAKER_01: Or if you're bringing out something that already exists, make it look new.
• 16:18
  SPEAKER_01: What we've done with proxy is kind of interesting in that we didn't make it look like anything that exists.
• 16:24
  SPEAKER_01: With a humanoid, you make it look like something we know, humans, and you can kind of project onto it what its capabilities are.
• 16:30
  SPEAKER_00: You mean more that it doesn't look like something that exists in nature versus it doesn't look like existing robots?
• 16:36
  SPEAKER_01: Maybe the folks at Mayo Clinic will disagree with me on this, but I think when when we first showed them proxy, they were interested in our consulting services, our flywheel services to help them think about how to bring robotics into the Mayo ecosystem.
• 16:50
  SPEAKER_01: But when we first showed them proxy, I looked around the room and I didn't feel like that many people fully got it.
• 16:59
  SPEAKER_01: It didn't look like anything that they had seen before.
• 17:02
  SPEAKER_01: It didn't look like a little two-wheel diff drive, didn't look like, you know, a lot of like Locust Fetch, those robots look similar.
• 17:08
  SPEAKER_01: A lot of AMRs are low to the ground, like Amazon's robots.
• 17:12
  SPEAKER_00: Let me interject right there.
• 17:13
  SPEAKER_00: So when speaking of proxy specifically, like this is gonna sound silly, but what's to get?
• 17:19
  SPEAKER_00: Like, what weren't they getting?
• 17:22
  SPEAKER_01: The use of the swerve drive, the use of four independently steerable wheels, the fact that this robot can do extremely precise positioning, like very, very precise positioning, it means that it can like come up to the edge of a table or go through a narrow passageway very, very precisely, right?
• 17:38
  SPEAKER_01: Or if a human's walking alongside, it can just like slide three inches to the right, like you or I would.
• 17:44
  SPEAKER_01: And so it makes the robot far more trustworthy in these environments.
• 17:49
  SPEAKER_01: If you have a two-wheel diff drive, um, in order for you to like move sideways slightly, you've got to make this like spline like curve, but it makes it look unnatural, it makes it look kind of drunken, it makes it look like it's wobbling.
• 18:02
  SPEAKER_01: If it has to correct itself, it's kind of swerving back and forth.
• 18:06
  SPEAKER_01: In fact, I you know, I've walked into these places that have you know fetch and locusts, and they the operators describe it as like, oh yeah, that one's drunk today, right?
• 18:14
  SPEAKER_01: And it's drunk because like its calibration is slightly off and it's just like constantly repositioning, so it's just moving a little bit.
• 18:21
  SPEAKER_00: And better, better the robot drunk than the forklift driver.
• 18:24
  SPEAKER_01: That's right.
• 18:25
  SPEAKER_01: But frankly, no one trusts things that look drunk in a in an operational environment.
• 18:30
  SPEAKER_01: And so they kind of get used to it, but they don't fully trust it.
• 18:33
  SPEAKER_01: And proxy doesn't have that problem, even if there's like an inch or two of calibration drift, it just kind of moves slightly to side, like like you or I would, but stays on path, right?
• 18:43
  SPEAKER_01: So that's one thing they didn't fully get the fact that it can just move existing carts, it's funny because you know, all these other ones there were like in conversations about can you attach to some other cart?
• 18:53
  SPEAKER_01: Do we need to modify the cart?
• 18:54
  SPEAKER_01: Do we need to bring your cart in?
• 18:56
  SPEAKER_01: Oh, the robot itself is a cart, but what if the cart, what if the robot's not used much because the carts don't move all the time?
• 19:02
  SPEAKER_01: Like, just that like realization that if you have a bunch of carts, like 90% of them are probably idle, right?
• 19:08
  SPEAKER_01: And so if you're putting all the sensing and autonomy on on an idle cart, that's a huge amount of cost.
• 19:14
  SPEAKER_01: Or if you can move the existing carts.
• 19:15
  SPEAKER_01: So again, these little things of like, hey, the positioning's really precise, the the sensing's really, really stable and secure because we see the world from six feet in the air.
• 19:27
  SPEAKER_01: Uh, we can move your existing carts so it's okay that 90% of your carts might not be moving at any one time.
• 19:32
  SPEAKER_01: We're not adding, you know, to the cost of that.
• 19:35
  SPEAKER_01: And and and people are like, yeah, swappable battery sounds great, until you've actually tried to run robotics at scale and you realize how much of a pain in the ass charging and charging stands are, right?
• 19:46
  SPEAKER_01: And so that that's kind of the benefit we had from Amazon was we knew precise positioning mattered.
• 19:51
  SPEAKER_01: We knew where you saw the world from mattered, we knew that carts were often idle and you need to be able to drop them off and leave them somewhere.
• 19:59
  SPEAKER_01: We knew that sitting on a charger was was terrible.
• 20:02
  SPEAKER_01: Like you think about sitting on a charger, one of the things people don't always realize is that you know circuit breakers are 15 amps usually, right?
• 20:10
  SPEAKER_01: Well, you you know, your charger, you have four or five robots charging and you're exceeding the 15 amps, right?
• 20:18
  SPEAKER_01: And so, where are you what you now have to like upgrade the electrical in the system in order to get enough amperage going to that part of the building to charge 20 or 30 robots?
• 20:29
  SPEAKER_01: It's and and so like once we went into Mayo and put the robots in their specimen processing lab, and people saw, and we these weren't the first robots they had tried in there.
• 20:42
  SPEAKER_01: People saw how well it worked.
• 20:43
  SPEAKER_01: Then the light bulb started to go off.
• 20:45
  SPEAKER_01: Then it was like, oh, this is this just works so much better, right?
• 20:50
  SPEAKER_01: And sometimes you just have to see it to realize that.
• 20:53
  SPEAKER_01: But to see, oh, this works so much better.
• 20:55
  SPEAKER_01: And now, like, I think it pretty much every week, another department at Mayo is saying, hey, can we can we think about getting these robots for this use case?
• 21:03
  SPEAKER_01: So the light bulb has kind of gone off like across the whole campus.
• 21:06
  SPEAKER_01: And now it's it's a little bit on us to like how quickly can we can we get robots uh going for them in all these process paths.
• 21:15
  SPEAKER_01: But but the point is when you're bringing something new to market that people haven't seen before, um it's not always immediately obvious, you know, why it's so much better.
• 21:26
  SPEAKER_01: Like I think about the you know, the first iPhone didn't look quite like a phone, right?
• 21:30
  SPEAKER_01: It looked very, very different.
• 21:32
  SPEAKER_01: And it took a little while for sales to really ramp for people to realize why that experience was so much better.
• 21:38
  SPEAKER_01: Um, but once they do, then the adoption just and you know, that's what we're excited about with COVID.
• 21:44
  SPEAKER_01: You know, the worry you have with robots is when you put them in the field, people start to say, Okay, well, this is this doesn't work as well here as I I hoped.
• 21:54
  SPEAKER_01: And and look, you know, we had things we were excited about at Amazon that didn't work as well as we hoped, and that's not great.
• 22:00
  SPEAKER_01: For the startup, right?
• 22:01
  SPEAKER_01: Because they put a lot of time and and capital into investing in that technology.
• 22:05
  SPEAKER_01: And if it if it doesn't work as well as you'd hope, then it can be kind of a dead end for a for a hardware company.
• 22:11
  SPEAKER_01: So we're really excited that like proxy has been working, frankly, better than we and the customers hoped at taking on these tasks.
• 22:19
  SPEAKER_01: And actually the folks in the lab at Mayo started to work with us on some ways to use it that we hadn't expected.
• 22:26
  SPEAKER_01: And that's always great when you start to realize you do have a platform, right?
• 22:30
  SPEAKER_01: You do have something that can be used in lots of different ways.
• 22:34
  SPEAKER_01: And you know, that's what we we discovered and learned with with uh Amazon's Hercules kind of Kiba Row.
• 22:39
  SPEAKER_01: We could turn them into drive sorters, we could we could modify that platform in really interesting ways.
• 22:45
  SPEAKER_01: Um, and so I think you're gonna see that with proxy too.
• 22:48
  SPEAKER_01: You're gonna see that you know, the manipulators we use, some of the ways that we interact with boxes, totes, carts, items, trays, is gonna evolve in in really interesting ways.
• 22:59
  SPEAKER_01: And we're gonna be able to do you know very, very cool things.
• 23:02
  SPEAKER_00: It's interesting because it it seems like there's two there's almost two opposing forces here.
• 23:07
  SPEAKER_00: It you know, it's obvious to me why whether a VC or you know, even especially uh this example of the Mayo clinic, somebody that really has to be pragmatic and practical about deploying these things because you know you can't shut the Mayo clinic down to to in to have to have new robots.
• 23:26
  SPEAKER_00: But there also seems to be this counterintuitive thing that's happening there as well, which is in a certain sense, it might be more difficult to sell small changes in in in the sense that like they want this like really radical thing that doesn't look like anything that exists.
• 23:44
  SPEAKER_00: But obviously, you know, there are there are a lot of ways in which proxy is is more like an AMR than a legged humanoid.
• 23:51
  SPEAKER_01: Sure.
• 23:51
  SPEAKER_01: Okay.
• 23:52
  SPEAKER_00: There's a difficulty selling like uh the big radical change, but you know, if it's trying to convince them that no, actually, like a wheeled robot is the right solution, I'm suspecting could be difficult, especially around kind of the you know, all of the conversations being had around humanoids right now.
• 24:07
  SPEAKER_01: For better or worse, the humanoids aren't working in these production environments.
• 24:11
  SPEAKER_01: So it's not uh, you know, um as much as this, you know, created some invent investor excitement.
• 24:18
  SPEAKER_01: There was, there has been some customer excitement about humanoids as well, but that customer excitement dissipates very quickly when people realize the technology is not available and not ready and doesn't work.
• 24:32
  SPEAKER_01: And I mean, again, this is why I get branded a skeptic because I sound harsh when I say that.
• 24:36
  SPEAKER_01: It's like, oh my god, he's saying, you know, Tesla said they're gonna, you know, deploy 10,000 next year.
• 24:41
  SPEAKER_01: Okay.
• 24:42
  SPEAKER_01: Well, great.
• 24:43
  SPEAKER_01: When that happens, that'll happen.
• 24:44
  SPEAKER_01: But like, that's not what anyone on the inside is saying is gonna happen, right?
• 24:48
  SPEAKER_01: And the deployments that we do have are fenced off, right?
• 24:52
  SPEAKER_01: They're not collaborative, so they're not working in spaces like like hospitals, and you know, the economics are still prohibitively expensive, so you know, it's maybe Tesla has deep enough pockets to go build a thousand of you know these quarter million plus dollar robots, but even Tesla seems to be maybe worried about cash right now because uh yeah, as I said, this is this is why I end up getting branded as a skeptic.
• 25:22
  SPEAKER_01: But I think for us, the good news is like our robot works.
• 25:25
  SPEAKER_01: It's it's running 16-hour day shifts at Marsk, it's running every day at Mayo, like it's doing the work, it's doing you know, uh pretty much 100% of the eligible work every day.
• 25:37
  SPEAKER_01: And so, you know, so maybe there's a little bit of a a little bit of a like trying to help people understand how the robot can help them.
• 25:47
  SPEAKER_01: But then when we do, then we can deliver the robots and put them out.
• 25:50
  SPEAKER_01: And we're, you know, we're gonna launch a couple new customers this this year, um, you know, talking to uh to our existing customers about you know much bigger expansions by the end of the year and next year.
• 26:03
  SPEAKER_01: And so I both think we have a better solution, and I think you know, we're we're well positioned to start to to deploy now, and and that's just gonna give us a a head start.
• 26:14
  SPEAKER_01: Um, I do I did talk to uh an innovation leader, big manufacturing company that you know seriously tested two humanoids in the past year, and they they said their conclusion was the tech is at least five years away.
• 26:29
  SPEAKER_01: Um and that was that was their assessment.
• 26:32
  SPEAKER_01: Uh and I had that same, you know, we were looking in 2018 and I was like, this feels like it's eight to ten years away, and and I was probably optimistic.
• 26:43
  SPEAKER_00: Obviously, it's not just skepticism, but it's it's also setting realistic expectations.
• 26:47
  SPEAKER_00: When you're talking about Elon Musk, you know, realistic expectations maybe aren't the first thing that comes to mind.
• 26:52
  SPEAKER_00: He's you know, he's had a history of overpromising some some things in in the past, but there's also a sense in which uh the the kind of overpromising is built into at least the way that robots are being sold.
• 27:08
  SPEAKER_00: You and I, a couple of years ago when I was at TechCrunch, um, we collaborated on a story about uh fudging robotic demos.
• 27:15
  SPEAKER_00: And I think that humans we we've seen a similar trend play out in LLMs where humans are sort of predisposed to think that these systems are smarter and more capable than they actually are.
• 27:28
  SPEAKER_01: I think there's an incredible amount of pressure to oversell your capability, right?
• 27:34
  SPEAKER_01: The because um you know what one of the rules I I've learned with operators, right, who are running warehouses, logistics, manufacturing is if a technology is more than a year away, then come back to me in a year.
• 27:51
  SPEAKER_01: It makes perfect sense from an operator standpoint.
• 27:53
  SPEAKER_01: If it's more than a year away, I've got things I gotta solve right now.
• 27:56
  SPEAKER_01: I've got plenty of problems that are, you know, on the urgent side.
• 27:59
  SPEAKER_01: And if it's more than a year away, I'll just park this until next year.
• 28:02
  SPEAKER_01: You kind of have to bring them solutions that can get in the field within a year, right?
• 28:07
  SPEAKER_01: Otherwise, they're not that interested.
• 28:10
  SPEAKER_01: But you need to show some interest to get investors interested, and investors don't want to hear that it's more than a year away either.
• 28:17
  SPEAKER_01: So there's this constant pressure to basically promise this things right around the corner.
• 28:21
  SPEAKER_01: Now, the game people play is to then, you know, look, uh Elon's been promising self-driving cars or Teslas are right around the corner every year, and it's because that is what people want to hear.
• 28:35
  SPEAKER_01: And I think what he's learned is that the penalty for under-delivering is maybe not as high as the reward for over-promising.
• 28:47
  SPEAKER_01: To Elon's credit, if you can over-promise and over-deliver just a few times, right, then you become credible.
• 28:56
  SPEAKER_01: And I think Elon has done that enough, right?
• 29:00
  SPEAKER_01: I mean, catching a rocket on chopsticks is freaking insane, right?
• 29:05
  SPEAKER_01: It is absolutely insane.
• 29:06
  SPEAKER_01: No one would have said that was possible on the timeline he did it.
• 29:10
  SPEAKER_01: Again, he's promised a lot of things that haven't come through, but he's promised some things where they've over-delivered.
• 29:16
  SPEAKER_01: And as a result, he's credible, right?
• 29:18
  SPEAKER_01: I mean, there's a low Earth orbit network of satellites that gives you high speed, relatively low latency internet everywhere right now.
• 29:26
  SPEAKER_01: Like it's insane.
• 29:27
  SPEAKER_01: It's great.
• 29:28
  SPEAKER_01: And and he he did that reasonably fast.
• 29:32
  SPEAKER_01: Again, Elon isn't isn't 100% an over-delivered, but he's over-delivered enough that he's credible.
• 29:38
  SPEAKER_01: So I think that's the tricky thing in robotics is you've got a bit over-promise and you've got to over-deliver.
• 29:45
  SPEAKER_01: Um, I I remember a funny story with Maris, because I knew I had to promise within a year, and uh Rez Agmoni, who's an amazing leader in technology innovation, he switched moved over to the venture capital world.
• 30:00
  SPEAKER_01: But he was excited about our technology very early and was willing to sign on very early and and had it.
• 30:07
  SPEAKER_01: He understood where it was going to be useful right away.
• 30:10
  SPEAKER_01: And I didn't have a robot.
• 30:12
  SPEAKER_01: I had sketches.
• 30:13
  SPEAKER_01: I'm not even sure when I pitched him.
• 30:15
  SPEAKER_01: I had a mechanical engineer yet.
• 30:17
  SPEAKER_01: He came to visit like nine months later, and I still didn't have a robot.
• 30:24
  SPEAKER_01: He'd worked with a lot of startups, he understood founders overpromising, but I think he gave me a little bit more credibility and respect because I came out of Amazon and big tech, and you know, maybe I'm not like you know, these founders who constantly overprom things and underdeliver.
• 30:37
  SPEAKER_01: I've since talked to him about it.
• 30:38
  SPEAKER_01: I said, when you came that June, you thought I was completely bullshitting and you didn't.
• 30:42
  SPEAKER_01: And he's like, Yeah, no, I was I was not happy.
• 30:46
  SPEAKER_01: Um and what we had decided to do was we realized that we couldn't incrementally build our way to the product, that we had to go into the design tools and just design the whole robot and build the first ones up from scratch.
• 31:01
  SPEAKER_01: It was a huge bet, right?
• 31:03
  SPEAKER_01: It was a huge, it was kind of like the NVIDIA bet where they just decided to do everything in simulator because they didn't have enough money to do it otherwise.
• 31:10
  SPEAKER_01: I knew we couldn't meet the merits timelines if we tried to progressively build.
• 31:14
  SPEAKER_01: So we'd stopped building and we started just designing everything and then ordering parts.
• 31:19
  SPEAKER_01: And sure enough, like a year later, maybe a year and a month, we might have been a little short.
• 31:25
  SPEAKER_01: We built the first full one full up.
• 31:28
  SPEAKER_01: Uh, I showed Arez the robot moving around and doing stuff, and then within a couple more months, we had it in his facility, and he was he was shocked.
• 31:37
  SPEAKER_01: So again, this is the like you earn some credibility because we did deliver.
• 31:42
  SPEAKER_01: I knew I had to promise that we were gonna get it out there, and then you had to find a way to deliver it.
• 31:46
  SPEAKER_01: And if you can do that and you can over-deliver, then I think you earn respect, you earn credibility, uh, and you earn that with investors as well.
• 31:54
  SPEAKER_01: And so that's our approach.
• 31:56
  SPEAKER_00: Yeah, I think there's another element of this too that maybe doesn't get discussed enough.
• 32:02
  SPEAKER_00: I think boss and dynamics played a role in this early on.
• 32:05
  SPEAKER_00: Obviously, a lot of their successes, you know, they were an early adopter of YouTube, and that's what got their name out there.
• 32:11
  SPEAKER_00: But now we're at this point, and I I don't think Cobot is in quite the same position, but all humanoid robot makers are where they feel like they have to sell to non-roboticist, normal people watching YouTube as much as they do to customers or investors.
• 32:28
  SPEAKER_01: We have an amazing marketing advisor.
• 32:30
  SPEAKER_01: He used to be the chief marketing officer for T-Mobile.
• 32:33
  SPEAKER_01: Uh, and he he's observed that with new technology, it does sometimes pay to play almost a more consumer style marketing campaign because you're trying to create awareness, you're trying to create momentum, you're trying to appeal to a more general audience.
• 32:54
  SPEAKER_01: It is true that I think you're trying to the humanoid comes through, and and in fairness, like we might not be doing it quite to the same degree, but you look at the industrial design of proxy, you look at the videos we put out, they're closer to kind of emotional content like Star Wars and those type of robots that are more mass appeal than they are to like the industrial AMR.
• 33:17
  SPEAKER_01: Like our robot would look odd in the AMR catalog, you know, industrial magazine that comes out.
• 33:22
  SPEAKER_01: But we do think that robots are going to go mainstream.
• 33:26
  SPEAKER_01: We're going to see them working in airports and hospitals and these places.
• 33:31
  SPEAKER_01: And customer adoption, mass market adoption, consumer appeal are going to matter.
• 33:38
  SPEAKER_01: So, yes, I would say we're all adopting a bit of a more consumer playbook.
• 33:43
  SPEAKER_01: While at the same time, if you look at you know, the conferences we're going to and the events and where our where our sales teams are, they're at these industrial forums, but all of us are trying to appeal to the public kind of enthusiasm for robotics.
• 33:58
  SPEAKER_00: Yeah, which I think is a perfectly fine thing if that means that these videos are going to be better lit or there's higher production values.
• 34:06
  SPEAKER_00: But I also worry, and I think you share this worry, that that also incentivizes faking in demos.
• 34:14
  SPEAKER_01: I think people have gotten better, right?
• 34:16
  SPEAKER_01: As the public has gotten more sophisticated about consuming the videos, some of the games people play go away.
• 34:24
  SPEAKER_01: This idea that we project human capabilities onto things with the human form factor.
• 34:28
  SPEAKER_01: If you think robots from movies, right, Transformers and Terminator, all of them are stronger than humans.
• 34:36
  SPEAKER_01: So we naturally think that a robot that looks like a human is stronger than a human.
• 34:40
  SPEAKER_01: We don't realize that it's like 60% the strength of a human.
• 34:44
  SPEAKER_01: Actually, it complicated things a little bit that Atlas, Boston Dynamics first humanoid, is pneumatically driven.
• 34:51
  SPEAKER_01: And pneumatic actuation is almost 10x the strength of electric motors.
• 34:56
  SPEAKER_01: The Atlas was a robot.
• 34:58
  SPEAKER_01: Atlas could almost do a backflip off its ankle motors.
• 35:01
  SPEAKER_01: So Atlas also helps set this impression that humanoids are incredibly strong.
• 35:05
  SPEAKER_01: And in practice, they're not.
• 35:08
  SPEAKER_01: But you can't get that from a video, really.
• 35:11
  SPEAKER_01: Sure.
• 35:12
  SPEAKER_01: But if you pay attention, if you know this and you go look at the video, like you notice that all the totes are empty, right?
• 35:18
  SPEAKER_01: That the robots are moving around, or that you know, that part that it moved from the automotive station to one other piece appears to be either carbon fiber or aluminum, not heavy steel, right?
• 35:32
  SPEAKER_01: And so the consumers of these videos have to become smarter in order not to be to be duped by that.
• 35:38
  SPEAKER_01: But it's almost impossible not to want to project human capability onto these things.
• 35:44
  SPEAKER_00: So there's that element of things.
• 35:46
  SPEAKER_00: There's there's the expectations when it comes to what the hardware can do, like as far as what the robot can lift.
• 35:52
  SPEAKER_00: But, and I'm gonna give you a perfect opportunity right here to plug your newsletter, um people similarly project these ideas onto what AI is capable of doing and how close we actually are to something that any reasonable person would uh define as AGI.
• 36:11
  SPEAKER_01: The advances in artificial intelligence have been really, really quite incredible, right?
• 36:15
  SPEAKER_01: The the fact that if you had asked me five years ago whether we would have a robot that could clear the dishes from a from a restaurant table or could understand semantically the concept of clearing the dishes from the table first, I would have thought that we would have been able to clear the table before we semantically understood what it meant to clear a table.
• 36:35
  SPEAKER_01: Now, LLM has clearly semantically understand what it means to clear a table and can provide a high-level robotic plan to lower level actuation to go do that.
• 36:44
  SPEAKER_01: We still don't have all the actuation to pick up a plate and not have the fork fall off, but we do have sophisticated AI that understands these semantic concepts and can transform them into action plans.
• 36:57
  SPEAKER_01: So the field of AI is is evolving very, very quickly.
• 37:01
  SPEAKER_01: I think what we know about AI is it's going to keep improving.
• 37:04
  SPEAKER_01: It's gonna improve a bit in step functions.
• 37:06
  SPEAKER_01: Um, it tends to improve as we get the right techniques against a very large data set and we kind of collapse it down into a model.
• 37:14
  SPEAKER_01: I think the challenge we have in in physical AI is we don't have the data sets.
• 37:19
  SPEAKER_01: And I also think that the way that we, you know, interact with things, right?
• 37:25
  SPEAKER_01: I give this example of like, how do you actually open this thing, right?
• 37:29
  SPEAKER_01: It doesn't, it's not obvious how it opens.
• 37:31
  SPEAKER_01: Um, and then when you pull it out, if you just grip it, it doesn't come out.
• 37:35
  SPEAKER_01: You learn by just playing that you put your thumb back here and then you transfer it, right?
• 37:41
  SPEAKER_01: And we don't even realize we did that, right?
• 37:44
  SPEAKER_01: It's not even a conscious thing.
• 37:47
  SPEAKER_01: We just kind of play with our hands, and our hands are just very, they have their own level of intelligence to them.
• 37:53
  SPEAKER_01: And I think it's a slightly different type of intelligence.
• 37:55
  SPEAKER_01: I've I've been calling it kind of the the equivalent of large language models where we're doing structured reasoning.
• 38:02
  SPEAKER_01: I think we play with our hands, we the way we reason.
• 38:05
  SPEAKER_01: We don't call it reason, we call it play.
• 38:07
  SPEAKER_01: We play, we play with that loose doorknob or we play with that stuck valve or whatever, right?
• 38:12
  SPEAKER_01: And uh, and it we call it play actually because it's rewarding.
• 38:16
  SPEAKER_01: Like trying to figure something out with your hands is kind of fun.
• 38:20
  SPEAKER_01: There's something to reinforcement learning, I think there's something to the self-reward, the kind of knowledge of what it feels like to get that valve unstuck that is gonna be the key to having dexterous manipulation that can play in the world.
• 38:39
  SPEAKER_01: And that's so that's that's what we're working on on the AI front.
• 38:42
  SPEAKER_01: I have written a piece on this recently on kind of physical AI and why uh uh you know, we talk about stochastic parrots, the idea that LLMs are just parroting back things they've heard before.
• 38:53
  SPEAKER_01: I think of it as stochastic puppets in robotics, the idea that these things are just being reanimated with the animations that they were were instructed before.
• 39:02
  SPEAKER_01: I don't think stochastic parroting has actually worked incredibly well.
• 39:07
  SPEAKER_01: Um, I am not as convinced that stochastic puppets uh are gonna be how we get to dexterous manipulation at scale.
• 39:17
  SPEAKER_01: And then, you know, thank you for mentioning the the newsletter.
• 39:20
  SPEAKER_01: I have just put out a newsletter.
• 39:21
  SPEAKER_01: I've been doing these Sunday videos just to share my thoughts on the space.
• 39:25
  SPEAKER_01: They've become incredibly popular.
• 39:27
  SPEAKER_01: I get a lot of comments on them.
• 39:29
  SPEAKER_01: I get a lot of people who feel like they understand me and how I'm thinking about better.
• 39:33
  SPEAKER_01: I did it a little bit not to just be perceived as a like curmudgeonly skeptic.
• 39:37
  SPEAKER_01: Like there's a lot more nuance and depth to how I think about this stuff.
• 39:40
  SPEAKER_01: And so trying to put longer form comment content out there.
• 39:44
  SPEAKER_01: Um, but now yeah, I'm trying to fan it out in a in a newsletter.
• 39:47
  SPEAKER_01: I'm like almost 4,500 subscriptions in the first week.
• 39:51
  SPEAKER_01: So that that's awesome.
• 39:52
  SPEAKER_01: And then we just took the the weekly content and put it out as a podcast this week as well, and so we'll keep doing that.
• 39:58
  SPEAKER_01: So I'm trying to just get the content out there in part because yeah, there's no value in being a skeptic in such an exciting new space.
• 40:06
  SPEAKER_01: There is a value to nuance and pragmatism and reality of what really works.
• 40:12
  SPEAKER_01: And so, in as much as I was getting sound bited down into a skeptic, I'm trying to get out there with a little longer form and kind of give some nuance.
• 40:21
  SPEAKER_00: I'm glad that you brought the AirPods out for the playing example, because as I was reading the newsletter on uh stochastic puppetry, and you threw out the word playing.
• 40:32
  SPEAKER_00: My mind, again, I'm creative writing major over here, uh UC Santa Cruz graduate over here.
• 40:38
  SPEAKER_00: Um, my mind automatically goes to like a robot solving a Rubik's Cube or playing soccer or these sorts of pastimes that we would um generally define as playing.
• 40:50
  SPEAKER_00: Obviously, like those do have value.
• 40:51
  SPEAKER_00: There's a reason why um roboticists and researchers are are operating in that space, but it it sounds like you've got a maybe squishier or at least like broader definition of what constitutes playing.
• 41:03
  SPEAKER_01: I do think of puzzles here.
• 41:05
  SPEAKER_01: I'll bring out another one that's uh that's exciting.
• 41:07
  SPEAKER_01: These are um apparently blacksmiths used to make these as like they're called tavern puzzles.
• 41:13
  SPEAKER_01: And so how do you get the ring off?
• 41:15
  SPEAKER_01: And it's it's really, really hard.
• 41:17
  SPEAKER_01: It took me like it took me about an hour every night for four nights to like figure out how to get the ring off of this.
• 41:22
  SPEAKER_01: And and and you start to like learn to reason in 3D and you develop a new way of anyhow.
• 41:28
  SPEAKER_01: That's a very sophisticated version.
• 41:30
  SPEAKER_01: This to me is like almost the Turing test of manipulation, right?
• 41:33
  SPEAKER_01: Like, if you can figure out how to solve a tavern puzzle with a with a robot, that is um that'll be incredibly impressive.
• 41:40
  SPEAKER_01: When I think though about like soccer, soccer is a great example.
• 41:43
  SPEAKER_01: Like learning to dribble a ball, bounce it from your you know, knees to your feet.
• 41:47
  SPEAKER_01: There is a rewarding thing like when you when you succeed in dribbling the ball 20 times, right?
• 41:53
  SPEAKER_01: Like without dropping it.
• 41:54
  SPEAKER_01: But there's this repetitive thing.
• 41:56
  SPEAKER_01: What we do is we we coach, we demonstrate, right?
• 42:01
  SPEAKER_01: So you teach a kid to swing a baseball bat, you're gonna demonstrate it a few times.
• 42:06
  SPEAKER_01: Then they're gonna try it, and they are not gonna do it very well necessarily the first couple times.
• 42:11
  SPEAKER_01: You're gonna give them a little bit of coaching because you can't do it for them, right?
• 42:15
  SPEAKER_01: You can't tell, you can't, it's very hard to puppet a kid, right?
• 42:19
  SPEAKER_01: We demonstrate for them and then we coach them a bit.
• 42:22
  SPEAKER_01: Hey, keep your elbow up, keep the bat in tighter, you know, pull your elbow forward.
• 42:27
  SPEAKER_01: Like we give them some verbal instruction, and then we let them practice.
• 42:32
  SPEAKER_01: I think there are two pieces where we're not sophisticated today.
• 42:35
  SPEAKER_01: I think the coaching piece, I think we can do a lot more in the coaching space in how do you give a small amount of guidance and correction, instruction to a robot without having to re-demonstrate it.
• 42:46
  SPEAKER_01: But then I think the play part is the play, it's that we have to make self-play interesting and rewarding.
• 42:53
  SPEAKER_01: And I hesitate to share more there because I think we're on to some things that we're quite excited about.
• 42:59
  SPEAKER_01: And so I don't really want to give too many hints about how we think that works.
• 43:04
  SPEAKER_01: But there's something to how we play that we need to imbue into robots.
• 43:10
  SPEAKER_00: So feel free to dodge or completely disregard this question.
• 43:13
  SPEAKER_00: But um, as as I was reading this stochastic puppetry piece, it seemed as though you were saying that play graduates into practice.
• 43:25
  SPEAKER_00: But as we're talking right now, it it seems like you're saying that practice is part of play.
• 43:30
  SPEAKER_00: So are these actually like two distinct ideas?
• 43:34
  SPEAKER_01: I I think there is a continuum there between practice and play.
• 43:37
  SPEAKER_01: Like to some extent, practice, and and we use those words semantically differently.
• 43:41
  SPEAKER_01: Practice means kind of repetitiveness of like learning a basic motion over and over again.
• 43:46
  SPEAKER_01: And I think of play as like a there's a bit of a discovery element to it, right?
• 43:50
  SPEAKER_01: There's a bit of a figuring things out.
• 43:53
  SPEAKER_01: Um and so, but when you're practicing for the first time, when you're when you're early on.
• 44:00
  SPEAKER_01: You're also discovering, right?
• 44:01
  SPEAKER_01: And so I think it's that discovery piece that's the exciting piece is the rewarding piece.
• 44:07
  SPEAKER_01: Um, and so I think there's there's a link between practice and play.
• 44:15
  SPEAKER_01: Um and but I I think of like, you know, playing with a doorknob that's a little stuck, right?
• 44:22
  SPEAKER_01: Like we're not gonna you can practice opening a doorknob over and over again, but you have to be able to come up to a new doorknob that's a little bit stuck and and wiggle it and figure it out and like and then figure out, oh, I've got to push the handle in slightly and then twist it and then it opens.
• 44:38
  SPEAKER_01: You don't want to have to use the whole demonstrate coach practice framework.
• 44:43
  SPEAKER_01: You want to get to the point where there's a there's an understanding of what it takes to open the door, what what opening the door feels like, um, what it looks like, what it feels like, what that experience is, and then you want the creativity and flexibility in your dexterity in order to in order to solve that.
• 45:03
  SPEAKER_01: We run into these little puzzles every day.
• 45:06
  SPEAKER_01: The refrigerator didn't open the way we expected it to.
• 45:09
  SPEAKER_01: Uh, we pulled something out of the refrigerator and something else started to fall, right?
• 45:14
  SPEAKER_01: And we we like very, very quickly solve these little physical puzzles all the time without that same level of like cognitive self-awareness.
• 45:23
  SPEAKER_01: We're not like talking to ourselves in our head.
• 45:25
  SPEAKER_01: There's no large language model, like in our head, pro maybe for some people there are.
• 45:29
  SPEAKER_01: For me, I'm not talking to myself as I as I like try to pull something out of a refrigerator.
• 45:34
  SPEAKER_01: I'm not narrating that activity.
• 45:36
  SPEAKER_01: Where when I'm trying to like solve uh a work-related problem or a customer problem, I'm often talking to myself in my head.
• 45:43
  SPEAKER_01: So physical, I think, is a just a different layer, a different way.
• 45:47
  SPEAKER_01: Um, and I think there's something really intrinsic to this idea of uh the kind of fun element of learning to do something new with your hands.
• 46:03
  SPEAKER_00: I know that Cobot, like you know, everybody else in the free world, has has a partnership with NVIDIA and is using some of that technology.
• 46:09
  SPEAKER_00: But something that really again jumped out at me in this piece was you said something along the lines of you know, simulation is going to be helpful, but the vast majority of the training is actually going to happen in person in the physical world.
• 46:24
  SPEAKER_00: And that that surprised me a bit.
• 46:27
  SPEAKER_01: You run into so many different scenarios that are novel, right?
• 46:31
  SPEAKER_01: I mean, you think about just like washing dishes, right?
• 46:35
  SPEAKER_01: And sure, we can kind of rotate train things, but understanding the texture of a dry sponge versus a wet sponge, understanding how to flip open the lid of a dish detergent because each one's a little bit different.
• 46:50
  SPEAKER_01: Like to some extent, we spend all this time gathering a lot of the world's knowledge and then we can push it back out as speech or text, right?
• 46:59
  SPEAKER_01: Um, in that form, I think we're a lot like I actually do think we're a lot like stochastic parrots, right?
• 47:08
  SPEAKER_01: But when I think about what you're doing every day, it's not it's not directly leveraging some past knowledge.
• 47:19
  SPEAKER_01: It is physically problem solving in this space of and people say, hey, we'll have this 3D world and we'll or we'll create these, you know, 3D world models, and then we'll be able to reason about, but that's not how we work either, right?
• 47:35
  SPEAKER_01: Yes, we roughly project the world into 3D, but you know, you can you can put us underwater with very different physics and dynamics, and we can still figure out how to deal with a you know, take a neutral buoyancy ball underwater, and like playing with that is very different than playing with a ball above the air, right?
• 47:58
  SPEAKER_01: And you know, when the first time you play with that underwater, you're not you don't have any prior knowledge of how to do it, and so you just play with it and you just figure out what its dynamics are, and then you figure out that you know, oh, this is kind of cool, and like, oh, a little torpedo rocket in water will go quite a bit further with just a little bit of and like you just discover these things.
• 48:22
  SPEAKER_01: I think we need to imbue robotics with that ability to kind of learn things through trial and error in the real world, and I think as we unlock that, we won't be playing around in simulation nearly as much.
• 48:38
  SPEAKER_01: The robots will just be trying new things in the world like you or I do.
• 48:43
  SPEAKER_00: Brad, always a pleasure.
• 48:45
  SPEAKER_00: Thank you so much for joining us.
• 48:46
  SPEAKER_01: Yeah, thank you, Brian.
• 48:47
  SPEAKER_01: This is great.
• 48:50
  SPEAKER_00: Thanks so much to Brad and everyone at Cobot who helped set up that conversation.
• 48:54
  SPEAKER_00: Check out Brad's top of mind newsletter and video podcast over on LinkedIn.
• 49:00
  SPEAKER_00: And you can find out more about what Cobot is up to over at their site.
• 49:04
  SPEAKER_00: It's co.bot.
• 49:07
  SPEAKER_00: As easy as possible to find that.
• 49:10
  SPEAKER_00: Don't forget to like and subscribe to the automated newsletter, uh, the podcast, the board game, the sequel to the motion picture.
• 49:18
  SPEAKER_00: You can find all of that at automated.fm.
• 49:21
  SPEAKER_00: Also fairly easy.
• 49:22
  SPEAKER_00: It's not co.bot, but you know, it's pretty straightforward.
• 49:26
  SPEAKER_00: And uh we will catch you next week with another episode of Automated.