#97 – Sertac Karaman: Robots That Fly and Robots That Drive

[0] The following is a conversation with Sirtash Karamon, a professor at MIT, co -founder of the autonomous vehicle company Optimus Ride, and is one of the top roboticists in the world, including robots that drive and robots that fly.

[1] To me, personally, he has been a mentor, a colleague, and a friend.

[2] He's one of the smartest and most generous people I know, so it was a pleasure and honor to finally sit down with him for this recorded conversation.

[3] This is the Artificial Intelligence Podcast.

[4] If you enjoy it, subscribe on YouTube, review it with five stars and Apple Podcasts, support on Patreon, or simply connect with me on Twitter at Lex Friedman, spelled F -R -I -D -M -A -N.

[5] As usual, I'll do a few minutes of ads now and never any ads in the middle that can break the flow of the conversation.

[6] I hope that works for you, and it doesn't hurt the listening experience.

[7] This show is presented by Cash App, the number one finance app in the App Store.

[8] When you get it, use the code Lex Podcast.

[9] Cash app lets you send money to friends, buy Bitcoin, and invest in the stock market with as little as $1.

[10] Since Cash App allows you to send and receive money digitally, let me mention a surprising fact about physical money.

[11] It costs $2 .4 to produce a single penny.

[12] In fact, I think it costs $85 million annually to produce them.

[13] That's a crazy little fact about physical money.

[14] So again, if you get Cash App from the App Store at Google Play and you use it, you the Code Lex podcast, you get $10, and Cash App will also donate $10 to first, an organization that is helping to advance robotics and STEM education for young people around the world.

[15] And now, here's my conversation with Sirtash Karamon.

[16] Since you have worked extensively on both, what is the more difficult task?

[17] Autonomous flying or autonomous driving?

[18] That's a good question.

[19] I think that autonomous flying, just kind of doing it for consumer drones and so on, the kinds of applications that we're looking at right now is probably easier.

[20] And so I think that that's maybe one of the reasons why it took off, like literally, a little earlier than the autonomous cars.

[21] But I think if we look ahead, I would think that, you know, the real benefits of autonomous flying, unleashing them in like transportation logistics and so on.

[22] I think it's a lot harder in autonomous driving.

[23] So I think my guess is that, you know, we've seen a few kind of machines fly here and there, but we really haven't yet seen any kind of, you know, machine.

[24] I at massive scale, large scale being deployed and flown and so on.

[25] And I think that's going to be after we kind of resolve some of the large scale deployments of autonomous driving.

[26] So what's the hard part?

[27] What's your intuition behind why at scale when consumer facing drones are tough?

[28] So I think in general, at scale is tough.

[29] Like for example, when you think about it, we have actually deployed a lot of robots in the, let's say, the past 50s.

[30] years.

[31] We's academics or we business entrepreneur?

[32] I think we as humanity.

[33] Humanity?

[34] A lot of people working on it.

[35] So we humans deployed a lot of robots.

[36] And I think that when you think about it, you know, robots, they're autonomous.

[37] They work.

[38] They work on their own.

[39] But they are either like an isolated environment or they are in sort of, you know, they may be at scale, but they are really confined to a certain environment that they don't interact so much with humans.

[40] And so, you know, they work in, I don't know, factory floors, warehouses, they work on Mars.

[41] You know, they are fully autonomous over there.

[42] But I think that the real challenge of our time is to take these vehicles and put them into places where humans are present.

[43] So now I know that there's a lot of like human -robot interaction type of things that need to be done and so on.

[44] That's one thing.

[45] but even just from the fundamental algorithms and systems and the business cases or maybe the business models, even like architecture, planning, societal issues, legal issues, there's a whole bunch of pack of things that are related to us putting robotic vehicles into human present environments.

[46] And these humans, you know, they will not potentially be even trained to interact with them.

[47] They may not even be using the services that are provided by these vehicles.

[48] they may not even know that they're autonomous.

[49] They're just doing their thing, living in environments that are designed for humans, not for robots.

[50] And that I think is one of the biggest challenges, I think, of our time to put vehicles there.

[51] And to go back to your question, I think doing that at scale, meaning, you know, you go out in a city and you have, you know, like thousands or tens of thousands of autonomous vehicles that are going around.

[52] It is so dense to the point where if you see one of them, you look around, you see another one.

[53] It is that dense.

[54] And that density, we've never done anything like that before.

[55] And I would bet that that kind of density will first happen with autonomous cars because I think we can ban the environment a little bit.

[56] We can, especially kind of making them safe is a lot easier when they're on the ground.

[57] When they're in the air, it's a little bit more complicated.

[58] But I don't see that there's going to be a big separation.

[59] think that, you know, there will come a time that we're going to quickly see these things unfold.

[60] Do you think there will be a time where there's tens of thousands of delivery drones that fill this guy?

[61] You know, I think it's possible, to be honest.

[62] Delivery drones is one thing, but, you know, you can imagine for transportation, like an important use case is, you know, we're in Boston, you want to go from Boston to New York, and you want to do it from the top of this building to the top of another building in Manhattan, and you're going to do it in one and a half hours.

[63] And And that's a big opportunity, I think.

[64] Personal transport.

[65] So like you and you be a friend, like almost like an Uber.

[66] So like four people, six people, eight people.

[67] In our work in autonomous vehicles, I see that.

[68] So there's kind of like a bit of a need for, you know, one person transport, but also like a few people.

[69] So you and I could take that trip together.

[70] We could have lunch.

[71] I think kind of sounds crazy, maybe even sounds a bit cheesy, but I think that those kinds of things are some of the real opportunities.

[72] And I think, you know, it's not like the typical airplane and the airport would disappear very quickly, but I would think that, you know, many people would feel like they would spend an extra $100 on doing that and cutting that four -hour travel down to one and a half hours.

[73] So how feasible are flying cars?

[74] It's been the dream.

[75] It's like when people imagine the future for 50 plus years, they think flying cars.

[76] It's like all technologies, it's cheesy to think about.

[77] now because it seems so far away, but overnight it can change.

[78] But just technically speaking in your view, how feasible is it to make that happen?

[79] I'll get to that question.

[80] But just one thing is that I think, you know, sometimes we think about what's going to happen in the next 50 years.

[81] It's just really hard to guess, right?

[82] Next 50 years, I don't know.

[83] I mean, we could get what's going to happen in transportation in the next 50 years.

[84] We could get flying saucers.

[85] I could bet on that.

[86] I think there's a 50 -50 chance that, you know, like you can build machines that can ionize the air around them and push it down with magnets and they would fly like a flying saucer.

[87] That is possible.

[88] And it might happen in the next 50 years.

[89] So it's a bit hard to guess, like when you think about 50 years before.

[90] But I would think that, you know, there's this kind of notion where there's a certain type of airspace that we call the agile airspace.

[91] And there's good amount of opportunities in that airspace.

[92] So that would be the space that is kind of a little bit higher than the place where you can throw a stone, because that's a tough thing when you think about it.

[93] You know, it takes a kid in a stone to take an aircraft down and then what happens.

[94] But, you know, imagine the airspace that's high enough so that you cannot throw a stone, but it is low enough that you're not interacting with the very large aircraft that are, you know, flying several thousand feet above.

[95] And that airspace is underutilized.

[96] Or it's actually kind of not utilized.

[97] Or it's actually kind of not utilize at all.

[98] Yeah, that's right.

[99] So there's, you know, there's like recreational people kind of fly every now and then, but it's very few.

[100] Like if you look up in the sky, you may not see any of them at any given time.

[101] Every now and now, you'll see one airplane kind of utilizing that space and you'll be surprised.

[102] And the moment you're outside of an airport a little bit, like it just kind of flies off and then it goes out.

[103] And I think utilizing that airspace, the technical challenge is there is, you know, building in autonomy and ensuring that that kind of autonomy is safe.

[104] Ultimately, I think it is going to be building in complex software or complicated so that it's maybe a few orders of magnitude more complicated than what we have on aircraft today and at the same time ensuring, just like we ensure on aircraft, ensuring that it's safe.

[105] And so that becomes like building that kind of complicated hardware and software becomes a challenge.

[106] especially when you know you build that hardware i mean you build that software with data and so um you know it's it's of course there's some rule -based software in there that kind of do a certain set of things but but then you know there's a lot of training there to do you think machine learning will be key to these kinds of delivering safe vehicles in the future uh especially flight not maybe the safe part but i think the intelligent part um i mean there are certain things that we do it with machine learning and it's just there's like right now no other way and and I don't know how else they could be done and you know there's there's always this conundrum I mean we could like could we like we could maybe gather billions of programmers humans who program perceptions that detect things in the sky and whatever or you know we I don't know we maybe even have robots like learn in a simulation environment and transfer.

[107] And they might be learning a lot better in a simulation environment than a billion humans put their brains together and try the program.

[108] Humans pretty limited.

[109] So what's the role of simulations with drones?

[110] You've done quite a bit of work there.

[111] How promising, just the very thing you said just now, how promising is the possibility of training and developing a safe flying robot in simulation and deploying it and having that work pretty well in the real world.

[112] I think that, you know, a lot of people, when they hear simulation, they will focus on training immediately, but I think one thing that you said, which was interesting, it's developing.

[113] I think simulation environments actually could be key and great for development.

[114] And that's not new.

[115] Like, for example, you know, there's people in the automotive industry have been using dynamic simulation for like decades now.

[116] And it's pretty standard that, you know, you would build and you would simulate.

[117] If you want to build an embedded controller, you plug that kind of embedded computer into another computer, that other computer would simulate tiny and so on.

[118] And I think, you know, fast forward these things, you can create pretty crazy simulation environments.

[119] Like, for instance, one of the things that has happened recently and that, you know, we can do now is that we can simulate cameras a lot better than we used to simulate them.

[120] We were able to simulate them before.

[121] And that's, I think we just hit the elbow on that kind of improvement.

[122] I would imagine that with improvements in hardware especially and with improvements in machine learning, I think that we would get to a point where we can simulate cameras very, very well.

[123] Simulate cameras means simulate how a real camera would see the real world.

[124] Therefore, you can explore the limitations of that.

[125] you can train perception algorithms on that in simulation, all that kind of stuff.

[126] Exactly.

[127] So, you know, it has been easier to simulate what we would call introspective sensors, like internal sensors.

[128] So, for example, inertial sensing has been easy to simulate.

[129] It has also been easily to simulate dynamics, like physics that are governed by ordinary differential equations.

[130] I mean, like how a car goes around, maybe how it rolls on the road, how it interacts with it interacts with the road, or even an aircraft flying around, like the dynamic of physics of that.

[131] What has been really hard has been to simulate extroceptive sensors, sensors that kind of like look out from the vehicle.

[132] And that's a new thing that's coming.

[133] Laser range finders that are a little bit easier.

[134] Cameras are a little bit tougher.

[135] I think once we nail that down, the next challenge, I think, in simulation, will be to simulate human behavior.

[136] that's also extremely hard even when you imagine how a human driven car would act around even that is hard but imagine trying to simulate a model of a human just doing a bunch of gestures and so on and you know it's actually simulated it's not captured like with motion capture but it is simulated that's very hard in fact today I get involved a lot with like sort of this kind of very high -end rendering projects and I have like this test that I pass it to my friends or my mom.

[137] You know, I send like two photos, two kind of pictures, and I say, rendered, which one is rendered, which one is real.

[138] And it's pretty hard to distinguish, except I realize, except when we put humans in there.

[139] It's possible that our brains are trained in a way that we recognize humans extremely well.

[140] But we don't so much recognize the built environments, because built environments sort of came after, per se, we evolved into sort of being humans, but humans were always there.

[141] Same thing happens, for example.

[142] you look at like monkeys and you can't distinguish one from another, but they sort of do.

[143] And it's very possible that they look at humans.

[144] It's kind of pretty hard to distinguish one from another, but we do.

[145] And so our eyes are pretty well trained to look at humans and understand if something is off, we will get it.

[146] We may not be able to pinpoint it.

[147] So in my typical friend test or mom test, what would happen is that we put like a human walking in a thing and they say, you know, this is not right.

[148] something is off in this video I don't know what but I can tell you it's the human I can take the human and I can show you like inside of a building or like an apartment and it will look like if we had time to render it it will look great and this should be no surprise a lot of movies that people are watching it's all computer generated you know even nowadays even you watch a drama movie and like there's nothing going on action wise but it turns out it's kind of like cheaper I guess to render the background and so they would but how do we get there?

[149] How do we get a human that would pass the mom slash friend test, a simulation of a human walking?

[150] Do you think that's something we can creep up to by just doing kind of a comparison learning where you have humans annotate what's more realistic and not just by watching?

[151] Like what's the path?

[152] Because it seems totally mysterious.

[153] how we simulate human behavior.

[154] It's hard because a lot of the other things that I mentioned to you, including simulating cameras, right?

[155] It is the thing there is that, you know, we know the physics.

[156] We know how it works, like in the real world.

[157] And we can write some rules and we can do that.

[158] Like, for example, simulating cameras, there's this thing called ray tracing.

[159] I mean, you literally just kind of imagine it's very similar to, it's not exactly the same, but it's very similar to tracing photon by photon.

[160] They're going around, bouncing on things and coming in your eye, a human behavior, developing a dynamic, like a model of that, that is mathematical so that you can put it into a processor that would go through that.

[161] That's going to be hard.

[162] And so what else do you got?

[163] You can collect data, right?

[164] And you can try to match the data.

[165] Or another thing that you can do is that, you know, you can show the front test.

[166] You know, you can say this or that.

[167] this or that, and that will be labeling.

[168] Anything that requires human labeling, ultimately, we're limited by the number of humans that, you know, we have available at our disposal and the things that they can do, you know, they have to do a lot of other things than also labeling this data.

[169] So that modeling human behavior part is, I think, going to realize it's very tough.

[170] And I think that also affects, you know, our development of autonomous vehicles.

[171] I see them self -driving as well.

[172] Like, you want to use, so you're building, building self -driving, you know, at the first time, like right after Urban Challenge, I think, everybody focused on localization, mapping and localization.

[173] You know, slam algorithms came in.

[174] Google was just doing that.

[175] And so building these HD maps, basically that's about knowing where you are.

[176] And then five years later in 2012, 2013 came the kind of code -in -code AI revolution, and that started telling us where everybody else is.

[177] But we're still missing what everybody else is going to do next.

[178] And so you want to know where you are.

[179] You want to know what everybody else is.

[180] Hopefully you know what you're going to do next.

[181] And then you want to predict what other people are going to do.

[182] And that last bit has been a real challenge.

[183] What do you think is the role your own of your, the ego vehicle, the robot, the you, the you, the you, the robotic you in controlling and having some control of how the future enrolls of what's going to happen in the future that seems to be a little bit ignored in trying to predict the future is how you yourself can affect that future by being either aggressive or less aggressive or signaling in some kind of way sort of this kind of game theoretic dance seems to be ignored for the moment it's yeah it's it's totally ignored i mean it's it's quite interesting actually like how we um how we interact with things versus we interact with humans.

[184] So if you see a vehicle that's completely empty and it's trying to do something, all of a sudden it becomes a thing.

[185] So you interact with like you interact with this table.

[186] And so you can throw your backpack or you can kick it, put your feet on it and things like that.

[187] But when it's a human, there's all kinds of ways of interacting with a human.

[188] So if, you know, like you and I are face -to -face, we're very civil, you know, we talk and we understand each other for the most part.

[189] We'll see, you just, that's time.

[190] You never know what's going to happen.

[191] But the thing is that, like, for example, you and I might interact through YouTube comments, and, you know, the conversation may go at a totally different angle.

[192] And so I think people kind of abusing these autonomous vehicles is a real issue in some sense.

[193] And so when you're an ego vehicle, you're trying to, you know, coordinate your way, make your way.

[194] It's actually kind of harder than being a human.

[195] You know, it's like, it's you, you not only need to be as smart as kind of humans are, but you also, you're a thing.

[196] So they're going to abuse you a little bit.

[197] So you need to make sure that you can get around and do something.

[198] So I, in general, believe in that sort of game theoretic aspects, I've actually personally have done, you know, quite a few papers, both on that kind of game theory and also like this, this kind of understanding people's social value orientation, for example.

[199] you know some people are aggressive some people not so much and and you know like a robot could understand that by just looking at how people drive and as they kind of come an approach you can actually understand like if someone is going to be aggressive or or not as a robot and you can make certain decisions well in terms of predicting what they're going to do the hard question is you as a robot should you be aggressive or not one face with an aggressive robot right now it seems like aggressive is a very dangerous thing to do because it's costly from a societal perspective, how you're perceived.

[200] People are not very accepting of aggressive robots in modern society.

[201] I think that's accurate.

[202] So that it really is.

[203] And so I'm not entirely sure like how to have to go about.

[204] But I know I know for a fact that how these robots interact with other people in there is going to be.

[205] And that interaction is always going to be there.

[206] I mean, you could be interacting with other vehicles or other just people kind of like walking around and like I said the moment there's like nobody in the seat it's like an empty thing just rolling off the street it becomes like no different than like any other thing that's not human and so so people and maybe abuse is the wrong word but you know people maybe rightfully even they they feel like you know this is a human present environment designed for humans to be and and they they kind of they want to own it.

[207] And then, you know, the robots, they would need to understand it and they would need to respond in a certain way.

[208] And I think that, you know, this actually opens up like quite a few interesting societal questions for us as we deploy, like we talk, robots at large scale.

[209] So what would happen when we try to deploy robots at large scale, I think, is that we can design systems in a way that they're very efficient or we can design them that they're very sustainable.

[210] But ultimately, the sustainability efficiency tradeoffs, like they're going to be right to in there.

[211] And we're going to have to make some choices.

[212] Like, we're not going to be able to just kind of put it aside.

[213] So, for example, we can be very aggressive and we can reduce transportation delays, increase capacity of transportation.

[214] Or, you know, we can be a lot nicer and allow other people to kind of, quote, unquote, own the environment and live in a nice place, and then efficiency will drop.

[215] So when you think about it, I think sustainability gets attached to energy consumption or environmental impact immediately.

[216] And those are, those are there, but like livability is another sustainability impact.

[217] So you create an environment that people want to live in.

[218] And if robots are going around being aggressive, you don't want to live in that environment, maybe.

[219] However, you should note that if you're not being aggressive, then you're probably taking up some delays in transportation and this and that.

[220] So you're always balancing that.

[221] And I think this choice has always been there in transportation, but I think the more autonomy comes in, the more explicit, the choice becomes.

[222] Yeah, and when it becomes explicit, then we can start to optimize it, and then we'll get to ask the very difficult societal questions of what do we value more efficiency or sustainability.

[223] It's kind of interesting.

[224] I think that will happen.

[225] I think we're going to have to like, I think that the interesting thing about, like, the whole autonomous vehicles question, I think, is also kind of, I think a lot of times, you know, we have focused on technology development like hundreds of years and, you know, the products somehow followed and then, you know, we got to make these choices and things like that.

[226] But this is a good time that, you know, we even think about, you know, autonomous taxi type of deployments and the systems that would evolve from there.

[227] And you realize the business models are different, the impact on architecture is different, urban planning.

[228] You get into like regulations.

[229] And then you get into like these issues that you didn't think about before, but like sustainability and ethics is like right in the middle of it.

[230] I mean, even testing autonomous vehicles, like, think about it.

[231] You're testing autonomous vehicles in human present environments.

[232] I mean, the risk may be very small, but still, you know, it's a, it's a, you know, strictly greater than zero risk that you're putting people into.

[233] And so then you have that innovation, you know, risk tradeoff that you're in that somewhere.

[234] And we understand that.

[235] And we understand that.

[236] pretty now, pretty well now is that if we don't test, at least the development will be slower.

[237] I mean, it doesn't mean that we're not going to be able to develop.

[238] I think it's going to be pretty hard, actually.

[239] Maybe we can.

[240] I don't know.

[241] But the thing is that those kinds of tradeoffs we already are making.

[242] And as these systems become more ubiquitous, I think those tradeoffs will just really hit.

[243] So you are one of the founders of Optimus Ride, Antonov's Vehicle Company.

[244] We'll talk about it.

[245] Let me, on that point, ask maybe good examples, keeping Optimus right out of this question, sort of exemplars of different strategies on the spectrum of innovation and safety or caution.

[246] So like Waymo, Google Self -Driving Car, Waymo represents maybe a more cautious approach.

[247] And then you have Tesla on the other side headed by Elon Musk that represents a more, however, which adjective you want to use, aggressive, innovative, I don't know.

[248] But what do you think about the difference in the two strategies in your view?

[249] What's more likely, what's needed and is more likely to succeed in the short term and in the long term?

[250] Definitely some sort of a balance is kind of the right way to go.

[251] I do think that the thing that is the most important is actually like an informed public.

[252] So I don't mind, you know, I personally, like if I were in some place, I wouldn't mind so much, like taking a certain amount of risk.

[253] Some other people might.

[254] And so I think the key is for people to be informed and so that they can, ideally, they can make a choice.

[255] in some cases that kind of choice making that unanimously is of course very hard but I don't think it's actually that hard to inform people so I think in one case like for example even the Tesla approach I don't know it's hard to judge how informed it is but it is somewhat informed I mean you know things kind of come out I think people know what they're taking and things like that and so on but I think the underlying I do think that these two companies are a little bit kind of representing, like, of course, one of them seems a bit safer, the other one, or, you know, whatever the adjective for that is, and the other one seems more aggressive or whatever the adjective for that is.

[256] But I think, you know, when you turn the tables, there are actually two other orthogonal dimensions that these two are focusing on.

[257] On the one hand, for Raymo, I can see that, you know, they, I think they a little bit see it as research as well.

[258] So they kind of, they don't, I'm not sure if they're like really interested in like an immediate product.

[259] You know, they talk about it.

[260] Sometimes there's some pressure to talk about it.

[261] So they kind of go for it.

[262] But I think, I think that they're thinking maybe in the back of their minds, maybe they don't put it this way.

[263] But I think they realize that we're building like a new engine.

[264] It's kind of like call it the AI engine or whatever that is.

[265] And, you know, an autonomous vehicles is a very interesting embodiment of that engine that allows you to understand where the ego vehicle is, the ego thing is, where everything else is, what everything else is going to do, and how do you react?

[266] How do you actually interact with humans the right way?

[267] How do you build these systems?

[268] And I think they want to know that.

[269] They want to understand that.

[270] And so they keep going and doing that.

[271] And so on the other dimension, Tesla is doing something interesting.

[272] I mean, I think that they have a good product.

[273] People use it.

[274] I think that, you know, like, it's not for me, but I can totally see people, people like it and people, I think they have a good product outside of automation.

[275] But I was just referring to the automation itself.

[276] I mean, you know, like it kind of drives itself.

[277] You still have to be kind of, you still have to pay attention to it, right?

[278] But, you know, people seem to use it.

[279] So it works for something.

[280] And so people, I think people are willing to pay for it.

[281] People are willing to buy it.

[282] I think it's one of the other reasons why people buy a Tesla car.

[283] Maybe one of those reasons is Elon Musk is the CEO and he seems like a visionary person.

[284] That's what people think.

[285] It seems like a visionary person.

[286] And so that adds like 5K to the value of the car.

[287] And then maybe another 5K is the autopilot.

[288] And, you know, it's useful.

[289] I mean, it's useful in the sense that like people are using it.

[290] And so I can see Tesla.

[291] Sure, of course, they want to be visionary.

[292] They want to kind of put out a certain approach and they may actually get there.

[293] But I think that there's also a primary benefit of doing all these updates and rolling it out because, you know, people pay for it.

[294] And it's, you know, it's basic, you know, demand, supply, market, and people like it.

[295] They're happy to pay another 5K, 10K for that novelty or whatever that is.

[296] And they use it.

[297] It's not like they get it and they try it a couple times.

[298] It's a novelty, but they use it a lot of the time.

[299] And so I think that's what Tesla is.

[300] It's actually pretty different.

[301] Like they are on pretty orthogonal dimensions of what kind of things that they're building.

[302] They are using the same AI engine.

[303] So it's very possible that, you know, they're both going to be sort of one day kind of using a similar almost like an internal internal combustion engine.

[304] It's a very bad metaphor.

[305] But similar internal combustion engine and maybe one of them is building like a car, the other one is building a truck or something.

[306] So ultimately the use case is very different.

[307] So you, like I said, are one of the final.

[308] of Optimus Ride, let's take a step back.

[309] It's one of the success stories in the autonomous vehicle space.

[310] It's a great autonomous vehicle company.

[311] Let's go from the very beginning.

[312] What does it take to start an autonomous vehicle company?

[313] How do you go from my idea to deploying vehicles like you are in a bunch of places, including New York?

[314] I would say that, I think that, you know, what happened to us is it was the following.

[315] I think we've realized a lot of kind of talk in the autonomous vehicle industry back in, like, 2014 even, when we wanted to kind of get started.

[316] And I don't know, like, I kind of, I would hear things like fully autonomous vehicles, two years from now, three years from now.

[317] I kind of never bought it.

[318] You know, I was a part of MIT's Urban Challenge Entry.

[319] It kind of like has an interesting history.

[320] So I did in college and in high school, sort of a lot of mathematically oriented work.

[321] And I think I kind of, you know, at some point, it kind of hit me. I wanted to build something.

[322] And so I came to MIT's mechanical engineering program.

[323] And I now realized, I think my advisor hired me because I could do like really good math.

[324] But I told him that, no, no, no, I want to work on that urban challenge car.

[325] I want to build the autonomous car.

[326] And I think that was kind of like a process where we really learned, I mean, what the challenges are and what kind of limitations are we up against.

[327] you know, like having the limitations of computers or understanding human behavior.

[328] There's so many of these things.

[329] And I think it just kind of didn't.

[330] And so we said, hey, you know, like, why don't we take a more like a market -based approach?

[331] So we focus on a certain kind of market and we build a system for that.

[332] What we're building is not so much of like an autonomous vehicle only, I would say.

[333] So we build full autonomy into the vehicles.

[334] But, you know, the way we kind of see it is that, we think that the approach should actually involve humans operating them, not just not sitting in the vehicle.

[335] And I think today, what we have is today, we have one person operate one vehicle, no matter what that vehicle.

[336] It could be a forklift, it could be a truck, it could be a car, whatever that is.

[337] And we want to go from that to 10 people operate 50 vehicles.

[338] How do we do that?

[339] you're referring to a world of maybe perhaps teleoperation.

[340] So can you just say what it means for 10?

[341] It might be confusing for people listening.

[342] What does it mean for 10 people to control 50 vehicles?

[343] That's a good point.

[344] I think it's a very deliberately didn't call it teleoperation.

[345] Because what people think then is that people think away from the vehicle sits a person, sees like maybe puts on goggles or something, VR, and drives the car.

[346] So that's not at all what we meet.

[347] But we mean the kind of intelligence whereby humans are in control, except in certain places, the vehicles can execute on their own.

[348] And so imagine like a room where people can see what the other vehicles are doing and everything.

[349] And, you know, there will be some people who are more like air traffic controllers, call them like AV controllers.

[350] And so these AV controllers would actually see kind of like a whole map.

[351] and they would understand where vehicles are really confident and where they kind of need a little bit more help.

[352] And the help shouldn't be for safety.

[353] Help should be for efficiency.

[354] Vehicles should be safe no matter what.

[355] If you had zero people, they could be very safe, but they'd be going five miles an hour.

[356] And so if you want them to go around 25 miles an hour, then you need people to come in.

[357] And for example, the vehicle come to an intersection and the vehicle can say, you know, I can wait, I can inch forward a little bit, show my intent, or I can turn left.

[358] And right now it's clear, I can turn, I know that, but before you give me to go, I won't.

[359] And so that's one example.

[360] It doesn't mean necessarily we're doing that, actually.

[361] I think if you go down all that much detail, that every intersection you're kind of expecting a person to press a button, then I don't think you'll get the efficiency benefits you want.

[362] You need to be able to kind of go around and be able to do that.

[363] these things, but I think you need people to be able to set high -level behavior to vehicles.

[364] That's the other thing with autonomous vehicles.

[365] You know, I think a lot of people kind of think about it as follows.

[366] I mean, this happens with technology a lot.

[367] You know, you think, all right, so I know about cars, and I heard robots.

[368] So I think how this is going to work out is that I'm going to buy a car, press a button, and it's going to drive itself.

[369] And when is that going to happen?

[370] And people kind of tend to think about it that way.

[371] But when you think about what really happens is that something comes in in a way that you didn't even expect.

[372] If asked, you might have said, I don't think I need that, or I don't think it should be that, and so on.

[373] And then that that becomes the next big thing, coding code.

[374] And so I think that this kind of different ways of humans operating vehicles could be really powerful.

[375] I think that sooner than later, we might open our eyes up to a world in which you go around, walk in a mall, and there's a bunch of security robots, they're exactly operated in this way.

[376] You go into a factory or a warehouse, there's a whole bunch of robots that are parted exactly in this way.

[377] You go to a, you go to the Brooklyn Navy Yard, you see a whole bunch of autonomous vehicles, Optimus Ride, and they're operated maybe in this way.

[378] But I think people kind of don't see that.

[379] I sincerely think that there's a possibility that we may almost see like a whole mushrooming of this technology in all kinds of places that we didn't expect before.

[380] And that may be the real surprise.

[381] And then one day, when your car actually drives itself, it may not be all that much of a surprise at all.

[382] Because you see it all the time.

[383] You interact with them.

[384] You take the Optimus ride.

[385] Hopefully, that's your choice.

[386] And then, you know, you hear a bunch of things.

[387] You go around.

[388] You interact with them.

[389] I don't know.

[390] Like you have a little delivery vehicle that goes around the sidewalks and delivers you things.

[391] And then you take it.

[392] It says, thank you.

[393] And then you get used to that.

[394] And one day, your car actually drives itself.

[395] and the regulation goes by and you know you can hit the button asleep and it wouldn't be a surprise at all i think that may be the real reality so there's going to be a bunch of applications that pop up around autonomous vehicles some some of which maybe many of which we don't expect at all so if we look at optimist ride what do you think you know the the viral application that the one that like really works for people in mobility what do you think optimist ride will connect with in the near future first?

[396] I think that the first places that I like to target, honestly, is like these places where transportation is required within an environment, like people typically call it geo -fense.

[397] So you can imagine like roughly two -mile by two -mile, could be bigger, could be smaller type of an environment.

[398] And there's a lot of these kinds of environments that are typically transportation deprived.

[399] The Brooklyn Navy art that we're in today, we're in a few different places, but that was the one that was last publicized.

[400] That's a good example.

[401] So there's not a lot of transportation there.

[402] And you wouldn't expect, like, I don't know, I think maybe operating an Uber there ends up being sort of a little too expensive.

[403] Or when you compare it with operating Uber elsewhere, that becomes the elsewhere becomes the priority, and these places become totally transportation deprived.

[404] And then what happens is that, you know, people drive into these places, and to go from point A to point B, inside this place within that day they use their cars and so we end up building more parking for them to for example take their cars and go to the lunch place and I think that one of the things that can be done is that you know you can put in efficient safe, sustainable transportation systems into these types of places first and I think that you know you could deliver mobility in an affordable way affordable, accessible you know sustainable way.

[405] But I think what also it enables is that this kind of effort, money, area land that we spend on parking, we could reclaim some of that.

[406] And that is on the order of like, even for a small environment, like two mile by two mile, it doesn't have to be smack in the middle of New York.

[407] I mean, anywhere else, you're talking tens of millions of dollars.

[408] If you're smack in the middle of New York, you're looking at billions of dollars of savings just by doing that.

[409] And that's the economic part of it.

[410] And there's a societal.

[411] part, right?

[412] I mean, just look around.

[413] I mean, um, the places that we live are like built for cars.

[414] It, it didn't look like this just like a hundred years ago.

[415] Like today, no one walks in the middle of the street.

[416] It's for cars.

[417] We, no one tells you that growing up, but you grow into that reality.

[418] And so sometimes they close the road.

[419] It happens here.

[420] You know, like the celebration, they close the road.

[421] Still people don't walk in the middle of the road.

[422] Like just walk in the middle, and people don't.

[423] But I think it has so much impact.

[424] the car in the space that we have.

[425] And I think we talked about sustainability, livability.

[426] I mean, ultimately, these kinds of places that parking spots at the very least could change into something more useful or maybe just like park areas, recreational.

[427] And so I think that's the first thing that that we're targeting.

[428] And I think that we're getting like a really good response, both from an economic, societal point of view, especially places that are a little bit forward looking.

[429] And like, for example, Brooklyn Navy Art, they have tenants.

[430] There's distinct hierarchical like new lab.

[431] It's kind of like an innovation center.

[432] There's a bunch of startups there.

[433] And so, you know, you get those kinds of people.

[434] And, you know, they're really interested in sort of making that environment more livable.

[435] And these kinds of solutions that Optimus Ride provides almost kind of comes in and becomes that.

[436] And many of these places that are transportation deprived, you know, they have, they actually rent shuttles.

[437] And so, you know, You know, you can ask anybody.

[438] The shuttle experience is like terrible.

[439] People hate shuttles.

[440] And I can tell you why.

[441] It's because, you know, like the driver is very expensive in a shuttle business.

[442] So what makes sense is to attach 20, 30 seats to a driver.

[443] And a lot of people have this misconception.

[444] They think that shuttles should be big.

[445] Sometimes we get that our optimist right.

[446] We tell them we're going to give you like four seeders, six seeders.

[447] And we get asked like, how about like 20 seeders?

[448] I'm like, you know, you don't need 20 seeders.

[449] you want to split up those seats so that they can travel faster and the transportation delays would go down.

[450] That's what you want.

[451] If you make it big, not only you will get delays in transportation, but you won't have an agile vehicle.

[452] It will take a long time to speed up, slow down, and so on.

[453] You need to climb up to the thing.

[454] So it's kind of like really hard to interact with.

[455] And scheduling too, perhaps when you have more smaller vehicles, it becomes closer to Uber where you can actually get a personal, I mean, just the.

[456] the logistics of getting the vehicle to you becomes easier when you have a giant shuttle.

[457] There's fewer of them, and it probably goes on a route, a specific route that is supposed to hit.

[458] And when you go on a specific route and all seats travel together versus, you know, you have a whole bunch of them, you can imagine the route you can still have, but you can imagine you split up the seats and instead of, you know, them traveling like, I don't know, a mile apart, They could be like, you know, half a mile apart if you split them into two.

[459] That basically would mean that your delays, when you go out, you won't wait for them for a long time.

[460] And that's one of the main reasons, or you don't have to climb up.

[461] The other thing is that I think if you split them up in a nice way, and if you can actually know where people are going to be somehow, you don't even need the app.

[462] A lot of people ask us the app.

[463] We say, why don't you just walk into the vehicle?

[464] How about you just walk into the vehicle?

[465] recognizes who you are and it gives you a bunch of options of places that you go and you just kind of go there.

[466] I mean, people kind of also internalize the apps.

[467] Everybody needs an app.

[468] It's like you don't need an app.

[469] You just walk into the thing.

[470] But I think one of the things that, you know, we really try to do is to take that shuttle experience that no one likes and tilt it into something that everybody loves.

[471] And so I think that's another important thing.

[472] I would like to say that carefully, just like teleoperation, like we don't do shuttles.

[473] You know, we really kind of of thinking of this as a system or a network that we're designing.

[474] But ultimately we go to places that would normally rent the shuttle service that people wouldn't like as much, and we want to tilt it into something that people love.

[475] So you've mentioned this actually earlier, but how many optimist ride vehicles do you think would be needed for any person in Boston or New York?

[476] If they step outside, there will be, this is like a mathematical question.

[477] there'll be two optimist ride vehicles within line of sight.

[478] Is that the right number to, well, at least one?

[479] Like, for example, that's the density.

[480] So, meaning that if you see one vehicle, you look around, you see another one, too.

[481] Imagine, like, you know, Tesla would tell you they collect a lot of data.

[482] Do you see that with Tesla?

[483] Like, you just walk around and you look around, you see Tesla?

[484] Probably not.

[485] Very specific areas of California, maybe.

[486] Maybe.

[487] You're right.

[488] Like, there's a couple zip codes.

[489] But I think that's kind of important because, you know, like maybe the couple zip codes.

[490] The one thing that we kind of depend on, and I'll get to your question in a second by now, like we're taking a lot of tangents today.

[491] Hell yes.

[492] And so I think that this is actually important.

[493] People call this data density or data velocity.

[494] So it's very good to collect data in a way that, you know, you see the same place so many times.

[495] Like you can drive 10 ,000 miles around the country or you drive 10 ,000 miles.

[496] in a confined environment, you'll see the same intersection hundreds of times.

[497] And when it comes to predicting what people are going to do in that specific intersection, you become really good at it.

[498] Versus if you draw on like 10 ,000 miles around the country, you've seen that only once.

[499] And so trying to predict what people do becomes hard.

[500] And I think that, you know, you said what is needed.

[501] It's tens of thousands of vehicles.

[502] You know, you really need to be like a specific fraction of vehicle.

[503] Like, for example, in good times, in Singapore, you can go and you can just grab a cat.

[504] And they are like, you know, 10 %, 20 % of traffic, those taxis.

[505] Ultimately, that's why you need to get to.

[506] So that, you know, you get to a certain place where you really, the benefits really kick off in like orders of magnitude type of a point.

[507] But once you get there, you actually get the benefits.

[508] And you can certainly carry people.

[509] I think that's one of the things.

[510] People really don't like to wait for themselves.

[511] But, for example, they can wait a lot more for the goods if they order something.

[512] Like, you're sitting at home and you want to wait half an hour.

[513] That sounds great.

[514] People will say it's great.

[515] You're going to take a cab.

[516] You're waiting half an hour.

[517] Like, that's crazy.

[518] You don't want to wait that much.

[519] But I think, you know, you can, I think, really get to a point where the system at peak times really focuses on kind of transporting humans around.

[520] And then it's really, it's a good fraction of traffic to the point where, you know, you go, you look around and there's something there and you just kind of basically get in there and it's already waiting for you or something like that and then you take it if you do it at that scale like today for instance Uber if you talk to a driver right I mean Uber takes a certain cut it's a small cut or you know drivers would argue that it's a large cut but you know it's it's when you look at the grand scheme of things most of that money that you pay Uber kind of goes to the driver.

[521] And if you talk to the driver, the driver will claim that most of it is their time.

[522] You know, it's not spent on gas, they think.

[523] It's not spent on the car, per se, as much.

[524] It's like their time.

[525] And if you didn't have a person driving, or if you're in a scenario where, you know, like, 0 .1 person is driving the car.

[526] A fraction of a person is kind of operating the car, because, you know, one operates several.

[527] If you're in that situation, you realize that the internal combustion engine type of cars are very inefficient.

[528] You know, we build them to go on highways, they pass crash tests, they're like really heavy.

[529] They really don't need to be like 25 times the weight of its passengers or, you know, like area -wise and so on.

[530] But if you get through those inefficiencies and if you really build like urban cars and things like that, I think the economics really starts to check out, like to the point where, I mean, I don't know, you may be able to get into a car and it may be like.

[531] less than a dollar to go from A to B. As long as you don't change your destination, you just pay 99 cents and go that.

[532] If you share it, if you take another stop somewhere, it becomes a lot better.

[533] You know, these kinds of things, at least for models, at least for mathematics and theory, they start to really check out.

[534] So I think it's really exciting what OptimusRard is doing in terms of it feels the most reachable, like it'll actually be here and have an impact.

[535] Yeah, that is the idea.

[536] And if, we contrast that, again, we'll go back to our old friends, Waymo and Tesla.

[537] So Waymo seems to have sort of technically similar approaches as Optimus Ride, but a different, they're not as interested as having impact today.

[538] They have a longer -term sort of investment.

[539] It's almost more of a research project still, meaning they're trying to solve, as far as I understand, maybe you can differentiate, but they seem to want to do more unrestricted movement, meaning move from A to B, where A to B is all over the place, versus Optimus Ride is really nicely geo -fenced and really sort of established mobility in a particular environment before you expand it.

[540] And then Tesla is like the complete opposite, which is, is the entirety of the world actually is going to be automated.

[541] Highway driving, urban driving, every kind of driving, you know, you kind of creep up to it by incrementally improving the capabilities of the autopilot system.

[542] So when you contrast all of these, and on top of that, let me throw a question that nobody likes, but is a timeline.

[543] When do you think each of these approaches, loosely speaking, nobody can predict the future.

[544] We'll see mass deployment.

[545] So Yoha Musk predicts the craziest approach is, I've heard figures like at the end of this year, right?

[546] So that's probably wildly inaccurate, but how wildly inaccurate is it?

[547] I mean, first thing to lay out like everybody else, it's really, it's really hard to guess.

[548] I mean, I don't know, I don't know where Tesla can look at or Elon Musk can look at and say, hey, you know, it's the end of this year.

[549] I mean, I don't know what you can look at.

[550] You know, even the data that, you know, I mean, if you look at the data, even kind of trying to extrapolate the end state without knowing what exactly is going to go, especially for like a machine learning approach.

[551] I mean, it's just kind of very hard to predict.

[552] But I do think the following does happen.

[553] I think a lot of people, you know what they do is that there's something that I called a couple times time dilation in technology prediction happens.

[554] Let me try to describe a little bit.

[555] There's a lot of things that are so far ahead.

[556] People think they're close.

[557] And there's a lot of things that are actually close.

[558] People think it's far ahead.

[559] People try to kind of look at a whole landscape of technology development.

[560] Admittedly, it's chaos.

[561] anything can happen in any order at any time.

[562] And there's a whole bunch of things in there.

[563] People take it, clamp it, and put it into the next three years.

[564] And so then what happens is that there's some things that maybe can happen by the end of the year or next year and so on.

[565] And they push that into like a few years ahead because it's just hard to explain.

[566] And there are things that are like, we're looking at 20 years more, maybe, you know, hopefully in my lifetime type of things.

[567] And because, you know, we don't know.

[568] I mean, we don't know how hard it is even.

[569] Like, that's a problem.

[570] We don't know, like, if some of these problems are actually AI complete.

[571] Like, we have no idea what's going on.

[572] And, you know, we take all of that and then we clump it and then we say three years from now.

[573] And then some of us are more optimistic.

[574] So they're shooting at the end of the year.

[575] And some of us are more realistic.

[576] They say, like, five years.

[577] But, you know, we all, I think, it's just hard to know.

[578] And I think trying to predict products ahead two, three years, it's hard to know in the following sense.

[579] You know, like we typically say, okay, this is a technology company, but sometimes really you're trying to build something where the technology does.

[580] Like there's a technology gap, you know, like and Tesla had that with electric vehicles.

[581] You know, like when they first started, they would look at a chart, much like a Moore's Law type of chart, and they would just kind of extrapolate that out.

[582] And they'd say, we want to be here.

[583] What's the technology to get that?

[584] We don't know.

[585] It goes like this.

[586] So it's probably just going to keep going.

[587] With AI that goes into the cars, we don't even have that.

[588] Like we can't, I mean, what can you quantify?

[589] Like, what kind of chart are you looking at, you know?

[590] But so I think when there's that technology gap, it's just kind of really hard to predict.

[591] So now I realize I talked like five minutes and avoid your question.

[592] I didn't tell you anything about that.

[593] It was very skillfully done.

[594] That was very well done.

[595] And I don't think you, I think you've actually argued that it's not a use, even any answer you provide now is not that useful.

[596] It's going to be very hard.

[597] There's one thing that I really believe in.

[598] And, you know, this is not my idea.

[599] And it's been, you know, discussed several times.

[600] But this, this kind of like something like a startup or a kind of an innovative company, including definitely maybe maybe even some of the other big companies that are kind of trying things.

[601] This kind of like iterated learning is very important.

[602] The fact that we're over there and we're trying things and so on, I think that's important.

[603] We try to understand.

[604] And I think that, you know, the code -and -code Silicon Valley has done that with business models pretty well.

[605] And now I think we're trying to get to do it where there's a literal technology gap.

[606] I mean, before, like, you know, you're trying to build.

[607] I'm not trying to, you know, I think these companies are building great technology to, for example, enable internet search to do it so quickly and that kind of didn't what wasn't there so much but at least like it was a kind of a technology that you could predict to some degree and so on and now we're just kind of trying to build you know things that it's kind of hard to quantify what kind of a metric are we looking at so psychologically as a sort of as a leader of graduate students and at optimist ride a bunch of brilliant engineers just curiosity psychological Do you think it's good to think that whatever technology gap we're talking about can be closed by the end of the year?

[608] Or do you, you know, because we don't know.

[609] So the way, do you want to say that everything is going to improve exponentially to yourself and to others around you as a leader?

[610] Or do you want to be more sort of maybe not cynical, but I don't want to use realistic.

[611] because it's hard to predict, but yeah, maybe more cynical, pessimistic about the ability to close that gap.

[612] Yeah, I think that, you know, going back, I think that iterated learning is, like, key, that, you know, you're out there, you're running experiments to learn.

[613] And that doesn't mean sort of like, you know, like you're optimist, right, you're kind of doing something, but, like, in an environment, but like what Tesla is doing, I think is also kind of like this kind of notion.

[614] And, you know, people can go around and say, like, you know, this year.

[615] next year, the other year, and so on.

[616] But I think that the nice thing about it is that they're out there, they're pushing this technology in.

[617] I think what they should do more of, I think, that kind of informed people about what kind of technology that they're providing, you know, the good and the bad and then, you know, and not just sort of, you know, if it works very well.

[618] But I think, I'm not saying they're not doing bad and informing.

[619] I think they're kind of trying.

[620] They, you know, they put up certain things or at the very least YouTube videos comes out on how the summon function works.

[621] everyone now and then.

[622] And, you know, people get informed.

[623] And so that kind of cycle continues.

[624] But, you know, I admire it.

[625] I think they kind of go out there and they do great things.

[626] They do their own kind of experiment.

[627] I think we do our own.

[628] And I think we're closing some similar technology gaps, but some also some are orthogonal as well.

[629] You know, I think like we talked about, you know, people being remote, like it's something or in the kind of environments that we're in or think about a Tesla car.

[630] Maybe you can enable it one day, like there there's low traffic, like you're kind of the stop on go motion, you just hit the button and you can really, or maybe there's another lane that you can pass into, you go in that.

[631] I think they can enable these kinds of, I believe it.

[632] And so I think that that part, that is really important and that is really key.

[633] And beyond that, I think, you know, when is it exactly going to happen and so on?

[634] I mean, it's like I said, it's very hard to predict.

[635] And I would imagine that it would be good to do some sort of like a one or two -year plan when it's a little bit more predictable that the technology gaps you close and the kind of sort of product that would ensue.

[636] So I know that from Optimus Ride or other companies that I get involved in.

[637] I mean, at some point you find yourself in a situation where you're trying to build a product and people are investing in that building effort.

[638] And those investors, they do want to know, as they compare the investments they want to make, they do want to know what happens in the next one or two years.

[639] And I think that's good to communicate that.

[640] But I think beyond that it becomes a vision that we want to get to someday and saying five years, 10 years, I don't think it means anything.

[641] But iterated learning is key, though.

[642] Do and learn.

[643] I think that is key.

[644] You know, I got to sort of throw back right at you, criticism in terms of, you know, like Tesla or somebody communicating, you know, how someone works and so on.

[645] I got the chance to visit Optimus Ride, and you guys are doing some awesome stuff, and yet the Internet doesn't know about it.

[646] So you should also communicate more showing off, you know, showing off some of the awesome stuff, the stuff that works and stuff that doesn't work.

[647] I mean, it's just the stuff I saw with the tracking of different objects and pedestrians, so I mean, incredible stuff going on there.

[648] Maybe it's just the nerd of me, but I think the world would love to see that kind of stuff.

[649] Yeah, that's well taken.

[650] I think, you know, I should say that it's not like, you know, we weren't able to.

[651] I think we made a decision at some point.

[652] That decision did involve me quite a bit on kind of sort of doing this in kind of code -in -code stealth mode for a bit.

[653] bit.

[654] But I think that, you know, we'll open it up quite a lot more.

[655] And I think that we are also at Optimus Ride kind of hitting a new new era.

[656] You know, we're big now.

[657] We're doing a lot of interesting things.

[658] And I think, you know, some of the deployments that we kind of announced were some of the first bits of information that we kind of put out into the world.

[659] We'll also put out our technology.

[660] A lot of the things that we've been developing is really amazing.

[661] And then, you know, We're going to start putting that out.

[662] We're especially interested in sort of like being able to work with the best people.

[663] And I think it's good to not just kind of show them when they come to our office for an interview, but just put it out there in terms of like, you know, get people excited about what we're doing.

[664] So on the autonomous vehicle space, let me ask one last question.

[665] So Elon Musk famously said that lighter is a crutch.

[666] So I've talked to a bunch of people about it.

[667] Got to ask you, you use that crutch quite a bit in the DARPA days.

[668] So, you know, and his idea in general, sort of, you know, more provocative and fun, I think, than a technical discussion.

[669] But the idea is that camera -based, primarily camera -based systems is going to be what defines the future of autonomous vehicles.

[670] So what do you think of this idea?

[671] Ladders are crutch versus primarily camera -based systems.

[672] first things first.

[673] I think, you know, I'm a big believer in just camera -based autonomous vehicle systems.

[674] Like, I think that, you know, you can put in a lot of autonomy and you can do great things.

[675] And it's, it's very possible that at the timescales, like I said, we can't predict 20 years from now.

[676] Like, you may be able to do things that we're doing today only with LIDAR and you may be able to do them just with cameras.

[677] And I think that, you know, you can can just, I think that I will put my name on it too.

[678] Like, you know, there will be a time when you can only use cameras and you'll be fine.

[679] At that time, though, it's very possible that, you know, you find the LiDAR system as another robustifier or it's so affordable that it's stupid not to, you know, just kind of put it there.

[680] And I think, and I think we may be looking at a future like that.

[681] Do you think we're over relying on LiDAR right now because we understand the better, it's more reliable in many ways in terms of from a safety perspective.

[682] It's easier to build with.

[683] That's the other thing.

[684] I think to be very frank with you, I mean, you know, we've seen a lot of sort of autonomous vehicles companies come and go and the approach has been, you know, you slap a LiDR on a car and it's kind of easy to build with when you have a LiDR, you know, you just kind of code it up and you hit the button and you do a demo.

[685] So I think there's, admittedly, there's a lot of people they focus on the LiDAR because it's easier to build with.

[686] That doesn't mean that, you know, without the camera, just cameras, you can, you cannot do what they're doing, but it's just kind of a lot harder.

[687] And so you need to have certain kind of expertise to exploit that.

[688] What we believe in and, you know, you maybe seeing some of it is that we believe in computer vision.

[689] We certainly work on computer vision at OptimusRide by a lot.

[690] And we've been doing that from day one.

[691] And we also believe in sensor fusion.

[692] So, you know, we do, we have a relatively minimal use of LIDARs, but we do use them.

[693] And I think, you know, in the future, I really believe that the following sequence of events may happen.

[694] First things first, number one, there may be a future in which, you know, there's like cars with LIDARs and everything and the cameras, but in this 50 -year -ahead future, they can just drive with cameras as well, especially in some isolated environments and cameras, they go and they do the thing.

[695] In the same future, it's very possible that, you know, the LIDARs are so cheap and frankly make the software maybe a little less compute -intensive at the very least, or maybe less complicated so that they can be certified or insured of their safety and things like that, that it's kind of stupid not to put the LIDAR.

[696] Like, imagine this.

[697] You either pay money for the LIDAR or you pay money for the compute.

[698] And if you don't put the LIDAR, it's a more expensive system because you have to put in a lot of compute.

[699] Like, this is another possibility.

[700] I do think that a lot of the sort of initial deployments of self -driving vehicles, I think they will involve LIDARs.

[701] And especially either low range or short, either short range or low -resolution LIDAR, are actually not that hard to build in solid state.

[702] They're still scanning, but like MAMS type of scanning LIDARs and things like that, they're like, they're actually not that hard.

[703] I think they will maybe kind of playing with the spectrum and the phase arrays.

[704] They're a little bit harder, but I think, like, you know, putting a MAM's mirror in there that kind of scans the environment.

[705] It's not hard.

[706] The only thing is that, you know, you just like with a lot of the things that we do nowadays in developing technology, you hit fundamental limits of the universe, the speed of flight becomes a problem in when you're trying to scan the environment, so you don't get either good resolution or you don't get range.

[707] But, you know, it's still, it's something that you can put in that affordably.

[708] So let me jump back to drones.

[709] You've, you have a role in the Lockheed Martin Alpha Pilot Innovation Challenge where teams compete in drone racing.

[710] It's super cool, super intense, interesting application of AI.

[711] So can you tell me about the very basics of the challenge and where you fit in, what your thoughts are on this problem?

[712] And it's sort of echoes of the early DARPA challenge through the desert that we're seeing now with drone racing.

[713] Yeah, I mean, one interesting thing about it is that, you know, people, the drone racing exists as an esport.

[714] And so it's much like you're playing a game, but there's a real drone going in an environment.

[715] A human being is controlling it with goggles on, so there's no, it is a robot, but there's no AI.

[716] There's no AI, yeah.

[717] Human being is controlling it, and so that's already there.

[718] And I've been interested in this problem for quite a while, actually, from a roboticist point of view, and that's what's happening in AlphaPilot.

[719] Which problem, of aggressive flight?

[720] Of aggressive flight.

[721] Fully autonomous, aggressive flight.

[722] The problem that I'm interested, I mean, you asked about Alpha Pilot, and I'll get there in a second, but the problem that I'm interested in, I'd love to build autonomous vehicles like drones that can go far faster than any human possibly can.

[723] I think we should recognize that we as humans have, you know, limitations in how fast we can process information.

[724] And those are some biological limitations.

[725] Like, we think about this AI this way, too.

[726] I mean, this has been discussed a lot, and this is not sort of my idea per se, but a lot of people kind of think about human level AI.

[727] And they think that, you know, AI is not human level.

[728] One day, it'll be human level and humans and AI, they kind of interact.

[729] Versus I think that the situation really is that humans are at a certain place and AI keeps improving and at some point it just crosses off and then, you know, it gets smarter and smarter and smarter.

[730] And so drone racing, the same issue.

[731] Humans play this game.

[732] And, you know, you have to like react a milliseconds.

[733] And there's really, you know, you see something with your eyes, and then that information just flows through your brain into your hands so that you can command it.

[734] And there's some also delays on, you know, getting information back and forth.

[735] But suppose those delays don't exist.

[736] You just, just a delay between your eye and your fingers is a delay that a robot doesn't have to have.

[737] So we end up building in my research group like systems that, you know, see things at a kilohertz, like a human and I would barely hit 100 hertz.

[738] So imagine things that see stuff in slow motion, like 10x slow motion.

[739] It will be very useful.

[740] Like we talked a lot about autonomous car, so, you know, we don't get to see it, but 100 lives are lost every day just in the United States on traffic accidents.

[741] And many of them are like known cases, you know, like you're coming through like a ramp, going into a highway, you hit somebody and you're off.

[742] Or, you know, like, you kind of get confused.

[743] You try to, like, swerve into the next lane.

[744] You go off the road and you crash, whatever.

[745] And I think if you had enough compute in a car and a very fast camera, right at the time of an accident, you could use all compute you have.

[746] Like, you could shut down the infotainment system and use that kind of computing resources.

[747] Instead of rendering, you use it for the kind of artificial intelligence that goes in there, the autonomy.

[748] And you can either take control of the company.

[749] and bring it to a full stop.

[750] But even if you can't do that, you can deliver what the human is trying to do.

[751] Human is trying to change the lane, but goes off the road, not being able to do that with motor skills and the eyes, and, you know, you can get in there.

[752] And there's so many other things that you can enable with what I would call high throughput computing.

[753] You know, data is coming in extremely fast, and in real time you have to process it.

[754] And the current CPUs, however fast you clock it, are typically not enough.

[755] You need to build those computers from the ground up so that they can ingest all that data.

[756] That I'm really interested in.

[757] Just on that point, just really quick, is the currently what's the bottom?

[758] Like you mentioned the delays in humans.

[759] Is it the hardware?

[760] So you work a lot with Nvidia hardware.

[761] Is it the hardware or is it the software?

[762] I think it's both.

[763] I think it's both.

[764] In fact, they need to be co -developed, I think, in the future.

[765] I mean, that's a little bit what Nvidia does.

[766] Sort of like they almost like build, the hardware and then they build the neural networks and then they build the hardware back and the neural networks back and it goes back and forth but it's that code design.

[767] And I think that you know like we try to way back.

[768] We try to build a fast drone that could use a camera image to like track what's moving in order to find where it is in the world.

[769] This typical sort of, you know, visual inertial state estimation problems that we would solve.

[770] And you know, we just kind of realize that we're at the limit sometimes of, you know, doing simple tasks.

[771] We're at the limit of the camera frame rate.

[772] Because, you know, if you really want to track things, you want the camera image to be 90 % kind of like, or somewhat the same from one frame to the next.

[773] And why are we at the limit of the camera frame rate?

[774] It's because camera captures data.

[775] It puts it into some serial connection.

[776] It could be USB or like there's something called camera serial interface that we use a lot.

[777] It puts into some serial connection.

[778] And copper wires can only transmit so much data.

[779] And you hit the Shannon limit on copper wires.

[780] And you know, you hit yet another kind of universal limit that you can transfer the data.

[781] So you have to be much more intelligent on how you capture those pixels.

[782] You can take compute and put it right next to the pixels.

[783] People are building those to do.

[784] How hard is it to get past the bottleneck of the copper wire?

[785] Yeah, you need to, you need to do a lot of parallel processing, as you can imagine.

[786] The same happens in the GPUs, you know, like the data is transferred in parallel somehow.

[787] It gets into some parallel processing.

[788] I think that, you know, like, now we're really kind of diverted off into so many different dimensions, but...

[789] Great.

[790] So it's aggressive flight.

[791] How do we make drones see many more frames a second in order to enable aggressive flight?

[792] That's a super interesting problem.

[793] That's an interesting problem.

[794] So, but like, think about it.

[795] You have, you have CPUs.

[796] You clock them at, you know, several gigahertz.

[797] We don't, clock them faster largely because, you know, we run into some heating issues and things like that.

[798] But another thing is that three gigahertz clock, light travels kind of like on the order of a few inches or an inch.

[799] That's the size of a chip.

[800] And so you pass a clock cycle and as the clock signal is going around in the chip, you pass another one.

[801] And so trying to coordinate that, the design of the complexity of the chip becomes so hard.

[802] I mean, we have hit the fundamental limits of the universe in so many things that we're designing.

[803] I don't know if people realize that.

[804] It's great, but we can't make transistors smaller because, like, quantum effects, the electrons start to tunnel around.

[805] We can't clock it faster.

[806] One of the reasons why is because, like, information doesn't travel faster in the universe.

[807] And we're limited by that.

[808] Same thing with the laser scanner.

[809] But so then it becomes clear that, you know, the way you organize the chip into a CPU or even a GPU, you now need to look at how to redesign that if you're going to stick with silicon.

[810] You could go do other things too.

[811] I mean, there's that too.

[812] But you really almost need to take those transistors, put them in a different way, so that the information travels on those transistors in a different way, in a much more way that is specific to the high -speed cameras coming in.

[813] And so that's one of the things that we talk about quite a bit.

[814] So drone racing kind of really makes that.

[815] embodies that.

[816] He embodies that.

[817] And it's exciting.

[818] It's exciting for people.

[819] You know, students like it.

[820] It embodies all those problems.

[821] But going back, we're building code and code another engine.

[822] And that engine, I hope one day will be just like how impactful seatbelts were in driving.

[823] I hope so.

[824] Or it could enable, you know, next generation autonomous air taxis and things like that.

[825] I mean, it sounds crazy, but one day we may need to perch land these things.

[826] If you really want to go from Boston to New York in one and a half hours, you may want to fix being aircraft.

[827] Most of these companies that are kind of doing flying cars, they're focusing on that.

[828] But then how do you land it on top of a building?

[829] You may need to pull off like kind of fast maneuvers for a robot, like perch landage, going to go and perch into a building.

[830] If you want to do that, like you need these kinds of systems.

[831] And so drone racing, you know, it's being able to go very faster.

[832] on any human can comprehend.

[833] Take an aircraft.

[834] Forget the quadcopter.

[835] You take a fixed wing.

[836] While you're at it, you might as well put some like rocket engines in the back and you just light it.

[837] You go through the gate and a human looks at it and just said, what just happened?

[838] And they would say it's impossible for me to do that.

[839] And that's closing the same technology gap that would, you know, one day steer cars out of accidents.

[840] So, but then let's get back to the practical, which is, sort of just getting the thing to work in a race environment, which is kind of what the, it's another kind of exciting thing, which the DARPA challenge to the desert did.

[841] You know, theoretically, we had autonomous vehicles, but making them successfully finish a race, first of all, which nobody finished the first year, and then the second year, just to get, you know, to finish and go at a reasonable time, is really difficult engineering, practically speaking, challenge.

[842] So that, let me ask about the, the alpha pilot challenge.

[843] There's a, I guess, a big prize potentially associated with the Bolognaic, reminiscent of the DARPA days, predictions.

[844] You think anybody will finish?

[845] Well, not soon.

[846] I think that depends on how you set up the race course.

[847] And so if the race course is a slow -on course, I think people will kind of do it.

[848] But can you set up some course, like literally some core?

[849] You get to design it as the algorithm.

[850] developer, can you set up some course so that you can beat the best human?

[851] When is that going to happen?

[852] That's not very easy.

[853] Even just setting up some course, if you let the human that you're competing with set up the course, it becomes a lot harder.

[854] So how many in the space of all possible courses would humans win and would machines win?

[855] Great question.

[856] Let's get to that.

[857] I want to answer your other question, which is like, the DARPA challenge days, right?

[858] What was really hard?

[859] I think we understood what we wanted to build, but still building things, that experimentation, that iterated learning, that takes up a lot of time, actually.

[860] And so in my group, for example, in order for us to be able to develop fast, we build like VR environments.

[861] We'll take an aircraft, we'll put it in a motion capture room, big, huge motion capture room, and we'll fly it.

[862] In real time, we'll render other images and beam it back to the drone.

[863] That sounds kind of notionally simple, but it's actually hard because now you're trying to fit all that data through the air into the drone.

[864] And so you need to do a few crazy things to make that happen.

[865] But once you do that, then at least you can try things.

[866] If you crash into something, you didn't actually crash.

[867] So it's like the whole drone is in VR.

[868] We can do augmented reality and so on.

[869] And so I think at some point, testing becomes very important.

[870] One of the nice things about AlphaPilot is that they built the drone, and they build a lot of drones, and it's okay to crash.

[871] In fact, I think maybe, you know, the viewers may kind of like to see things that's right.

[872] That potentially could be the most exciting part.

[873] It could be the exciting part.

[874] And I think, you know, as an engineer, it's a very different situation to be in.

[875] Like in academia, a lot of my colleagues who are actually in this race and they're really great researchers, but I've seen them trying to do similar things whereby they built this one drone, and, you know, somebody with like a face mask and a glows are going, you know, right behind the drone, trying to hold it if it falls down.

[876] Imagine you don't have to do that.

[877] I think that's one of the nice things about Alpha Pilot Challenge where, you know, we have these drones and we're going to design the courses in a way that we'll keep pushing people up until the crashes start to happen.

[878] And, you know, we'll hopefully sort of, I don't think you want to tell people crashing is okay.

[879] Like we want to be careful here, but because, you know, we don't.

[880] people to crash a lot.

[881] But certainly, you know, we want, we want them to push it so that, you know, everybody crashes once or twice.

[882] And, and, you know, they're really pushing it to their limits.

[883] That's where iterated learning comes in.

[884] Every crash is a lesson.

[885] There's a lesson, exactly.

[886] So in terms of the space of possible courses, how do you think about it?

[887] In the war of human versus machines, where do machines win?

[888] We look at that quite a bit.

[889] I mean, I think that, you know, you will see quickly that, like you can design a course and you know in in certain courses like in the middle somewhere if you kind of run through the course once you know the machine gets beaten pretty much consistently by slightly but if you go through the course like 10 times humans get beaten very slightly but consistently so humans at some point you know you get confused you get tired and things like that versus this machine is just executing the same line of code tirelessly, just going back to the beginning and doing the same thing exactly.

[890] I think that kind of thing happens.

[891] And I realize sort of as humans, there's the classical things, you know, that everybody has realized.

[892] Like if you put in some sort of like strategic thinking, that's a little bit harder for machines that I think you sort of comprehend, precision is easy to do.

[893] So that's what they excel in.

[894] And also, sort of repeatability is easy to do.

[895] That's what they excel in.

[896] You can build machines that excel in strategy as well and beat humans that way too, but that's a lot harder to build.

[897] I have a million more questions, but in the interest of time, last question.

[898] What is the most beautiful idea you've come across in robotics?

[899] Whether a simple equation, experiment, a demo, simulation, piece of software, what just gives you pause?

[900] That's an interesting question.

[901] I have done a lot of work myself in decision -making, so I've been interested in that area.

[902] So in robotics, you have somehow the field has split into, like, you know, there's people who would work on, like, perception, how robots perceive the environment, then how do you actually make, like, decisions?

[903] And there's people also, like, how to interact, people interact with robots, there's a whole bunch of different fields.

[904] And, you know, I have admittedly worked a lot on the more control and decision -making than the others.

[905] And I think that, you know, the one equation that has always kind of baffled me is Belmont's equation.

[906] And so it's this person who have realized, like, way back, you know, more than half a century ago on, like, how do you actually sit down?

[907] and if you have several variables that you're kind of jointly trying to determine, how do you determine that?

[908] And there is one beautiful equation that, you know, like today, people do reinforce them and we still use it.

[909] And it's baffling to me because it both kind of tells you the simplicity because it's a single equation that anyone can write down.

[910] You can teach it in the first course on decision making.

[911] At the same time, it tells you how computation we have.

[912] how hard the problem is.

[913] I feel like a lot of the things that I've done at MIT for research has been kind of just this fight against computational efficiency.

[914] Things, like, how can we get it faster?

[915] To the point where we now got to, like, let's just redesign this chip.

[916] Like, maybe that's the way.

[917] But I think it talks about how computationally hard certain problems can be by nowadays what people call curse of dimensionality.

[918] And so as the number of variables, kind of grow, the number of decisions you can make grows rapidly.

[919] Like, if you have, you know, 100 variables, each one of them take 10 values, all possible assignments is more than the number of atoms in the universe.

[920] It's just crazy.

[921] And that kind of thinking is just embodied in that one equation that I really like.

[922] And the beautiful balance between it being theoretically optimal and somehow, practically speaking, given the curse of dimensionality, nevertheless, in practice, works among, you know, despite all those challenges, which is quite incredible.

[923] Which is quite incredible.

[924] So, you know, I would say that it's kind of like quite baffling, actually, you know, in a lot of fields that we think about how little we know.

[925] You know, like, and so I think here too, you know, we know that in the worst case, things are pretty hard, but, you know, in practice, generally things work.

[926] So it's just kind of, kind of, it's kind of baffling decision making how little we know.

[927] Just like how little we know about the beginning of time, how little we know about, you know, our own future.

[928] Like, if you actually go into like from Belmont's equation all the way down, I mean, there's also how little we know about like mathematics.

[929] I mean, we don't even know if the axioms are like consistent.

[930] It's just crazy.

[931] Yeah.

[932] I think a good lesson there, just like as you said, we tend to focus on the worst case or the boundaries of everything we're studying.

[933] And then the average case seems to somehow work out.

[934] If you think about life in general, we mess it up a bunch, you know, we freak out about a bunch of the traumatic stuff.

[935] But in the end, it seems to work out okay.

[936] Yeah, it seems like a good metaphor.

[937] Sir Tash, thank you so much for being a friend, a colleague, a mentor.

[938] I really appreciate it.

[939] It's an honor to talk to you.

[940] Likewise.

[941] Thank you, Lex.

[942] Thanks for listening to this conversation with Sir Dash Karamon, and thank you to our presenting sponsor Cash App.

[943] Please consider supporting the podcast by downloading Cash app and using code Lex Podcast.

[944] If you enjoy this podcast, subscribe on YouTube, review it with five stars on Apple Podcast, support it on Patreon, or simply connect with me on Twitter at Lex Friedman.

[945] And now, let me leave you with some words from Hal 9 ,000, from the movie 2001 A Space Odyssey.

[946] I'm putting myself to the fullest possibly use, which is all I think that any conscious entity can ever hope to do.

[947] Thank you for listening and hope to see you next time.