Apptronik has a totally different approach to building humanoid robots

yeah so five foot eight weighs 160 pounds it can lift 55 pounds and it has a four hour battery initially and it's swappable so we're targeting 22 hours a day seven days a week up time it can also be tethered as well and opportunity charge like what you see in autonomous mobile robots [Music] who will win the race to have the world's first usable general purpose humanoid robot hello and welcome to Tech first my name is John gets here I thought I knew all the companies that are making general purpose robots right there's Tesla there's Sanctuary AI there's figure AI Fourier intelligence agility robotics Boston Dynamics I'm sure there's a bunch more that I don't know but I know that I was wrong because one that just popped up that launched this week is out-tronic they're based in Austin Texas they're partnering with NASA I want to hear more about that and they're building Apollo 5 foot 8 160 pound robot computer chat is CEO Jeff Cardenas welcome Jeff thanks for having me here there's so much competition right now what is going on it's exciting moment I think for robotics in in the world as a whole uh We've finally reached an inflection point and it's it's funny because when we first got started everyone told us not to do humanoids and now everyone's getting into the into the race but it's been exciting and interesting to see it all play out everyone is doing humanoids and that's a real challenge right I mean there there's pieces that that are known and capable shoulders necks maybe right um walking with a bipedal robot uh not necessarily the easiest thing in the world but it's been done and it can be done um really really challenging Parts obviously in the brain and the hands and fingers will get into all that tell us about Apollo which you've called the world's most capable humanoid robot yeah so Apollo is the result of many years of hard work and research and development at uptronic we've built 13 different uh robots overall and eight iterations on humanoids so we started in humanoids when we were still working in the lab at the University of Texas at Austin started working with NASA back then for the dark robotics challenge two of my co-founders Dr Nick Payne and Dr Luis dentist we're on the valkyrie team and basically uptronic was created in 2016 to commercialize the work out of NASA so back then Valkyrie was millions of dollars it was 300 pounds but it was one of the very first electric humanoid robots so we felt like Hey general purpose robots more versatile robots are going to be the future electric is going to be the way to go and we want to build a commercial version of this and so got it out into uh the world now finally seven years later uh in in many years of work it's really it's great to hear that because you launched with Apollo yesterday right and so kind of the world wakes up and says oh what's this new company what are they doing right and it's like that old phrase like an instant success you just weren't around to see all the work that went into it right um so you built a bunch of robots already you you built a number of iterations on the humanoid robot talk about that Journey a little bit yeah I mean for us we always saw that this was really a technology problem more than was a market problem I think a lot of you know entrepreneurs and other folks are looking to get into this space because they see the market opportunity but for many years the technology problems had to be solved to make it viable so it's interesting to hear you say you know walking we can do that that was not the case even five years ago we had no idea how to do Dynamic walking Boston Dynamics was really 10 years ahead of Academia in terms of the type of walking that they were doing and everybody was trying to figure out how to catch up and do what they were doing for many years and so you know piece by piece we had to solve these problems and you know the way that we viewed it was this is a technical Challenge and we need to solve the key pieces that are needed to make this real how do we go towards viable commercial general purpose robots and we basically just broke the problem down down into you know in from and solved it from first principles so started with electric actuation for humanoid robots we've done over 35 iterations on electric actuators some of those are small medium large of the same family of actuators but a tremendous amount of r d there elon's talked about the need for actuators for these robots and that's been our body of work that was my co-founder Dr Nick Payne that was his thesis in grad school was next generation actuation for legged robots um and you know the the electronics didn't exist we needed more real-time communication because you have a lot more sensing in these robots and then certainly the software which I'm sure we'll get into but we we started basically at the foundation and we bootstrapped the company so I mentioned it's funny that everyone's into humanoids now because when we got started everyone told us do not build Hardware focus on the software focus on the AI and the problem was the robots didn't exist we're like well what are we going to put this AI on eventually as as it as it matures and develops we don't have the robots yet so we have to build these systems and so we bootstrapped ourselves and what we would do is we would work for other companies we've worked with several big Automotive companies looking at things like humanoid robots and we've helped them build their systems we've built and delivered a variety of systems including some of the robots for Sanctuary we've partnered with sanctuary in the early days and uh have they've been a great you know partner along and we built their first prototypes for them and each time that we would we would build these robots we would iterate we would learn something new all getting towards a point of building the robot we always ultimately wanted to build which is Apollo and so you know the thing that I love about robotics is um you got at the end of the day you can talk about what a robot's going to do but you have to show it in the real world so our philosophy has always been Show versus tell so we didn't have a need to really get out there and say hey we're gonna do this our view was like well let's do it and then we'll show off what we do and we'll let our work speak for itself and so we really had our heads down over these years just trying to get this stuff working solving the technology problems um iterating pretty quickly as well and you know we've had robots walking for seven years we've iterated on we've built full systems in three months um so it's not that we've been you know taking our time doing this we've actually been cracking these problems and we're at this point where we've met the threshold where everything is good enough which I kind of think of this like the personal computer in 1982 right it's like it's the beginning of you know a lot of things had to build on each other and converged for this breakout moment to happen uh but we've got that work behind us and I think the robot we put out there um we're really proud of and excited to see where it goes I love that approach I really love that approach because you didn't put a guy in a suit and walk on a stage and said here's our robot all right and that has a couple hundred year history I'm glad you didn't do that um super interesting in terms of the approach talk about before we get into the details of the robot give us the high level specs I mentioned five foot eight 160 pounds um I think it's five hour battery life is it a swaffle battery is it a hook it up in charge how fast can it go what can it do yeah so five foot eight weighs 160 pounds it can lift 55 pounds and it has a four hour battery initially and it's swappable so we're targeting 22 hours a day seven days a week up time it can also be tethered as well and opportunity charge like what you see in autonomous mobile robots um it is fully electrically actuated we as I mentioned we've had a tremendous amount of iteration in that space over the years and we think that we've got a really unique solution for performance and cost right performance at cost so there's a trade-off there where you're you know you can purely focus on performance which to me you know Atlas is an amazing machine Boston Dynamics robot kind of like a Formula One car it's really performance optimized uh would be very difficult to mass manufacture Atlas you know it's got custom Hydraulics and all those things in it and so what we've really focused on is how do we get performance at Cost how do we find the right trade-off and ultimately build a commercial product that we can build for less than fifty thousand dollars is our Target and it can still have the performance that's needed to do the work that we needed to do and then through covid we learned a lot about the supply chain and so a lot of the ideas that are now in Apollo are getting around supply chain constraints so that we can really you know scale this thing up into big volumes and have we don't have any single Source vendors in terms of what it does initially we're focused on what we call Gross manipulation and that's compared to dexterous manipulation and we've learned that because we've built a lot of robots over the years and dexterous manipulation is very difficult we have a ton of respect for folks that are going after that in this space but it's a really difficult problem and the exciting thing for us at this juncture is we don't have to solve that problem to get these things out into the world turns out there's a huge shortage in in logistics and a lot of those tasks are just moving boxes or totes from point A to point B and that's something that we know how to do now and so that's where we're going to start but then the beauty of a general purpose robot is it's a software update away from doing something new and so we'll continue to get you know more advanced as we move uh we move into this and get them out into the world super interesting to talk about hands and gross manipulation versus dexterous uh one robot robotic CEO that I chatted with before said you know the robot's basically a hand delivery mechanism because the hands do all the work right and and he said actually creating a hand a robotic hand like the human hand in capability is beyond our capability right now it's not just beyond any one company it's beyond human capability right now there might be some different opinions on that uh we'll see but that's where of course maybe half of the degrees of freedom that a robot might have uh exists so it's a really challenging thing and then of course wear and tear on all those little Motors and the hens skin whatever you use for skin super super hard so I see that challenge you talked about a software update which is amazing that's incredible is there a possibility for like a hardware update so let's say three years from now you crack human hands maybe not quite as good as this because you want to hit your fifty thousand dollar Target level but you know ninety percent seventy percent and whatever enough for many manufacturing type jobs can you like take a hand off and plug a new hand in yeah yeah so Apollo is modular there's this big debate of Wheels versus legs too from sort of the traditional folks in automation like what do you need legs for um you know we can use Wheels in all these applications and and what we've done with Apollo is just taken everything we've learned because we've been building these robots with customers over many years and so we're able to take all that learning and inject it into Apollo and some people are going to want these things on Wheels there's a big there's a huge number of advantages to legs we think legs will win the day overall but there's this problem with legs that the robots can fall over and and so in some cases you can have wheels the beauty of robots is you can have your cake and you can eat it too and that you could build them to be modular so Apollo is modular at the Torso so if you want to put it on Wheels you can throw it on Wheels we think that will demonstrate that legs will be the most versatile um you know sort of platform long term and the reason we know about the challenges of wheelbases is because we've designed them and so we've deployed versions of humanoids on Wheels and we've learned from that and so the same thing is true with the end of vectors I I agree that's I'm sure that's Jordy uh that's saying hands and you know I agree that long term the humanoid needs hands but in the near term there's many applications where you don't require a full five fingered hand and I mean you can do things with a one degree of Freedom hand there's a whole range of things you can do there's a whole range of things that robots do today with pincher grippers or you know and and so you can you can sort of expand and you can you don't have to solve all the problems at once I have a ton of respect for Jordy I've worked with Jordy over many years he's you know he's a Visionary um but we've just taken a different approach in terms of how we've thought about that and we want to partner with folks like Sanctuary they've been a big partner on it uh with us you know they can put their hands on a robot like Apollo and and we can work together there as they sort of crack the dexterous manipulation problem so yeah it's a it's modular at the chest it's modular at the um at the end effectors it's also modular at the head as well in terms of putting different sensor payloads on it so we have a standard sort of camera based Vision system but there's also debates about lidar do you need lidar or not our vision approach doesn't need lidar but in some cases when you start to put these robots Outdoors if you want to add lidar you can this is something I think Boston Dynamics has done a really good job of with spot is they created the ability to put different mission payloads on the back of spot and something that we learned from along the way and part of what's designed into Apollo love it and as you hinted Jordy is uh Jordy Rose he's a CEO of sanctuary Ai and the former CEO of a Quantum Computing company that sold a 50 million dollar quantum computer to Google yeah and it's still around I want to there's so many places to go here I do want to talk about the brain that's really challenging right I mean you you gotta how are you building intelligence into your robots is is it is it pre-program Maneuvers is it versatility with a certain level of intelligence talk about how you're doing that yeah so I mean I think the long-term goal is to start to get towards more and more intelligence overall but I think in terms of AI and intelligence as a whole for humanoids you can really break it down into two buckets so the first bucket is physical intelligence so that's like coordination hand-eye coordination uh the ability to balance walking is part of that that's the physical intelligence the other side is cognitive intelligence so how do you make decisions how do you reason about the world how do you abstract ideas you know things like that what I would say is that we've really focused on building from the bottom up and there's different approaches you can go from the top down you know as in start with the intelligence and think about how to build a machine around that we've kind of gone from the bottoms up which is start with the actuators the motor controllers the electronics you know really the basic building blocks and then build up into intelligence my view is that you want to build the most capable platform you can possibly build and then you can think of these intelligences as software that you can put on top of the robot there's people that disagree with that and say well in order to get to full intelligence you need deeper integration and I think we'll see but we've really focused on this physical intelligence and the exciting thing for me and you had a question that maybe will get too deeper but is kind of like where are we at I kind of think of this as really an evolution of what's already being done out in the world and we don't have to solve new problems to get humanoids out into initial applications to show utility for humanoids to be fully realized yes you need you know much higher levels of intelligence than we have today to for the fully realized uh humanoid I think but we have a lot of the building blocks already today so for example we we in you know if you think of the evolution of Robotics 2004 collaborative robots came out those were human safe robots by 2010 compute got good enough batteries got good enough we could have mobile robots we started to do things like slam and navigation by 2016 machine learning came to the scene and we could do intelligent grasping so what we've done is sort of build on all these things that we've seen work in production and really build from the bottom up and integrate those things together and take in maybe more a conservative approach than some people are taking to basically you know use what we know works and use what we've known you know we can we can deploy into the world today and then you know we can always add these other sort of uh more difficult sort of r d problems later on down the road um and so you know we can dive in deeper where we want to go there it's fascinating to see the different approaches and that's the beauty of the sort of free market Innovation System that we operate in in the Western World at least where you do have those people are coming top down they want to build intelligence and intelligence will do everything that's a risky bit I mean it's an amazing bet if you make it and you win because if you win you've solved everything quote unquote everything right but if you don't win you end up with an expensive boondoggle that doesn't accomplish anything you know it doesn't just it's either it's either really good because you can't have a robot out in the wild maybe making a sound which are slicing something up or using a tool it could be dangerous that is potentially dangerous your approach in software mirrors your approach in Hardware which is starting from the ground up what can I do what do I know I can do what do I know I can do today and that seems to be a very very pragmatic and practical way of doing it I think that's super interesting I do want to go deeper um into what this means and what it looks like but maybe before we do that um you're doing this in Austin Texas and you feel that that is significant why I think if you if you look at you know there's people out there that are saying there's going to be more humanoid robots Than People one day um you know like I said it's funny to me to to have that out there when everyone thought these things weren't viable even five years ago that there were Novelties and I think that's exciting I'm not sure that they'll all be humanoids but I think there will be a lot of humanoids and I think it's the most versatile platform you can build um but the reason I think Texas is important is the question is where are we going to get all the robots we need so if you look at the world today you know there's hundreds of humanoids maybe right there's not very many of these systems so how are we going to go from hundreds to thousands to Millions to maybe billions and I believe that Mexico is going to play a big role in that for North America I think if you look at Mexico relative to other countries where we're doing manufacturing today it's got a lot of advantages it's about a third cost of Labor depending on how you measure it um it's geographically located you know uh really close to the U.S market so we can get from Monterey to Texas in two and a half hours anywhere in the U.S in 24 hours and they have the skill sets to be able to pull this off and so I've always felt like Texas Mexico Corridor was going to be one of the most important manufacturing corridors in the world over the coming decades and this is something that I was saying you know coming out of grad school you know we were we had two key ideas and we started uptronic one was that robots had to become more versatile and two was that somebody had to be building you know robots domestically here to serve the U.S market we didn't have any major domestic oems when we got started and so if you agree with the premise that hey we're going to need domestic manufacturers long term the question is where's that going to happen and typically it's been on the east or the West Coast the big hubs for robotics have been California or Boston but I think Texas actually has a number of unique advantages over those hubs where I really believed that Texas has actually much better adjacencies than those other um uh you know places and so I always felt like Texas was the place to do it we're already producing gearboxes and a lot of the heavy machinery we're doing a lot of that for the energy industry before but as you look towards what's coming next is there's a lot of people in Texas that are looking for okay as the oil and gas boom ends which it will at some point where do we apply all of this industrial base that we already have built and for a number of reasons I think robotics actually makes a ton of sense for that and um and you know that's why I think Texas is going to be important well it is interesting cost of living is certainly less proximity to Mexico I was going to make a joke you know what is this access to labor that you're talking about aren't the robots going to build the robots but I'm sure there's there's going to be some tricky jobs for humans and all that stuff let's look forward a little ways and and let's say we're in a future where we have tens of millions you know and and and and we're progressing how does this change our world has it changed our economy I think that it fundamentally changes the way that we live and work and the reason I think that is because as humans our most valuable resource is time and our time here is limited and you know you had great thinkers in the last century you know uh John Maynard Keynes famously predicted that we would have a 15-hour work week and so what I think changes is that instead of doing things that we have to do that somebody's just got to do um we can now have machines that do that for us and what that does is free us up to spend time on things that we really value you know why do we spend more time at work than we do with our families you know and the answer to that today is well someone's got to do that we've got to keep the economy running and going we have to provide goods and services to our fellow man uh in order to keep all this moving but I think what robotics has the potential to change is to change that equation what if the cost of goods and services dramatically Falls because they basically will you know slope towards the cost of the raw materials as the cost of Labor continues to go down so goods and services could become much cheaper and much more abundant than they are today and that frees us up to do and spend time the way that we want and there's this interesting quote that I heard but you know what did Darwin Galileo um Newton what do they all have in common they were all very wealthy and so they had time to think and contemplate their existence and think about these higher level ideas and today you have people that are stuck in a cycle of you know working all the time to make ends meet and um I think you know an optimistic version of the future is you start to Free People up they're able to think about things like their own health about taking care of each other we start to fix the health care of the education system and and ultimately we evolved as humans and I think that you know applied in the right way could be a really positive thing super interesting okay so you've launched the robot you've launched Apollo uh can what can somebody buy or rent today So today we're working on pilots and you know my whole philosophy of core sort of value for us at abtronic is Show versus tell so what we've done is we've built a demo Center here on site at obtronic we're mocking up the use cases that we're looking to deploy Apollo into so for the remainder of this year we're basically signing up pilot customers and we have some of the you know the Marquee customers in the world that have already signed up and we're doing these on-site proof of Concepts uh through the rest of this year where we're demonstrating it my what I tell the partners we work with if I can't do it here at my facility I can't do it at your facility so make sure that you're comfortable with the performance of the robots here and then we'll get it on site uh early next year so next year we start with um the initial sort of fielded pilots and we feel that a lot of systems up into this point so we've already put one of the things I tell people when they come to apptronic they're looking for all the robots and we have a lot now but for many years every robot we ever made we sold because that's how we stayed alive that was our you know that was our business model we were entirely funded on revenue for the first five years now we only just raised money in the last couple years and so um but next year we'll get them out into the the world we're not look we're not you know uh putting out pricing just yet but a big part of what we've cracked is the ability to make these things affordable and so that's going to be a big part of our value proposition as we move ahead you appear to have been remarkably Capital efficient um I know many of the other entities companies departments that are building or trying to build general general purpose robots they've raised a hundred million dollars they've they're part of a trillion dollar company or a multi-billion dollar company you've been Scrappy you've been bootstrapping you recently raised what 14 15 million dollars which is interesting but that all seems to Accord with Connie your philosophy of start small build they'll know what we know and even to to your go to market strategy of bringing people in and seeing what can do super super interesting and I I really like that actually at fifty thousand dollars assuming you can get there because that's your goal it's not there yet no you're not in Mass manufacturing yet I'm assuming either but assuming it's fifty thousand dollars that's a very interesting price point because if you look at the kinds of jobs that you're going to place Us in in logistics and stuff like that you're spending probably a bit more than that if you look at entire cost for having an employee and benefits and other things like that and that sounds interesting do you know how you're gonna bring them to Market are you going to sell them out right or are you going to have a SAS solution what's your thinking there yeah so you know once again the way that I ask right robots as a service not software as a service yeah robots as a service and and the one of the exciting things about why this is feasible now is we already have business models that exist for selling mobile robots it didn't exist before the AMR Market took off but now there's you know lots of companies that are selling mobile robots to these exact same markets in logistics and so customers now know how to buy them the two ways they're buying amr's today are either robots as a service or capex typically with a SAS component to that so some the larger companies want to own their own fleets they want volume discounts and other things and so they'll buy them outright but I think largely what you'll see in the early stages of the humanoid Market is a robot says a service model where you know they want to try them out they want to see how this works there's people that are worried about you know Tech like technical obsolescence right like how quickly is this going to mature and develop them are ready to buy a fleet yet or do I want to wait some time and so for a number of reasons I think robots as a service is going to you know be important for this and now as an industry we know how to do that there's third party financing groups that have already set up around this it's now um you know something that the market understands which wasn't the case you know five years ago I keep thinking about additional questions we're having a great conversation I I want to I want to ask this one where do you situate the United States and maybe North America a little more broadly in terms of global innovation of humanoid robots because you're obviously us-based uh there's a bunch of others figure we've talked about um the one in Vancouver right Jordy Rose's company um we've talked about Boston Dynamics right um uh so yeah Sanctuary is the one in in Vancouver I visited I think it was called robot Island it was the sort of the the the name for it in Denmark it's got to be about three years now just pre-covet uh where there's been there has been for about a couple decades kind of a global concentration of companies in Automation and and Robotics is odensa uh Denmark and I haven't seen anything come out of there and they've been instrumental with some of the big manufacturing companies globally with large-scale robots and automation stuff but I haven't seen a humanoid robot come out of there why am I seeing so many in the states and where do you situate the U.S in terms of global Innovation for humanoid general purpose robots yeah so I mean this is a topic that um is important to me because I think there is the race is on um there's a lot of interesting stuff that's coming out you know all over the world I think one of the reasons that you see the US leading right now is because the US government has invested so much money in the r d that was required to make this happen so whether it's figure Boston Dynamics or us many folks you know have teams that have their roots in something called the DARPA robotics challenge the DRC is you know they injected you know tens of millions I think it was over 100 million total um DARPA did to really Advance state of the art for general purpose robots and that was 2013 to 2015. and the seeds were planted back then and we're just now seeing the fruit that's being beared off of that investment but that's really what made Boston Dynamics big you know was DARPA funding in the early years Atlas came about for the DRC so the very first atlases were designed for the DARPA robotics Challenge and and that's really what spurred a lot of the innovation there has not been as much government funding in this space since the dark robotics Challenge and I think that if the U.S wants to continue to lead the government's going to need to step it in a big way and really inject more money into it but this is the same thing that happened with autonomous driving there was something called the DARPA Urban Challenge and there was a couple DARPA challenges that that really seeded the technology that moved it from the lab and give it the first nudge out into the commercial space and then companies were built out of that and so you know Jerry Pratt that's over at figure at ihmc he you know had a big role in the dark robotics Challenge and did great work there and then Boston Dynamics certainly uh came out of that in terms of who's leading in the world and where do we go from here I think China is making a big push in humanoids in particular and the government there is really stepping up to make it happen and so so I I really want to see the US government respond this is a race that I think is important long term you know how should these be deployed and I think it's an area that that we can lead there's other great countries doing amazing things as well there's great work happening in Korea certainly in Japan as well and then all across Europe the stuff coming out of eth Zurich and and groups like antibiotics they're doing really great things though not in humanoids in more versatile you know kind of Cutting Edge Next Generation robots it's important it is really important hey because if you win here you kind of win I mean we talked about the cost of information approaching zero we've talked about the cost of software uh approaching zero because replication is essentially free right if you achieve a working and capable general purpose humanoid robot the cost of Labor as we've already talked about starts to approach zero as well and all of a sudden that opens up a ton of capability for manufacturing cheaply onshore manufacturing other things you want to do jobs that were not you couldn't pay for them before maybe environmental Reclamation maybe Public Works projects you couldn't pay for them before you couldn't afford them before all of a sudden they become uh affordable and desirable and and everything that a society wants to as well if you are concerned about declining population in some of the older nations of the world Europe Japan all sorts of places China as well um those that's all critically important yeah this and I have an interesting story about that because the U.S actually invented the very first industrial robot so it was invented in the late 50s went into a General Motors Factory in the early 60s it was called the unimate arm and the company that built it unimation actually ended up folding in the in the 80s I've heard a variety of the you know different stories on what happened but long story short they didn't you know they didn't keep getting funded they didn't keep getting backed uh in a critical time General Motors I think was involved somehow and pulled out well so the U.S invented the very first industrial robot and effectively lost the first wave of Industrial Automation the big four that were producing all the industrial arms two were Japanese FANUC and Yaskawa one was Swiss ABB and the other one was German Kuka and so you know I think this is important for policy makers and others because this next wave is is dwarfs the first wave in impact and sized and so I think it's important to get it right um but it's an interesting story and something that I try to tell everyone that will listen when whenever I get a chance Jeff this has been a wonderful conversation thank you for taking the time yeah thank you very much

Channel: John Koetsier

Views: 51,252

Keywords: John Koetsier

Id: ab9QIzYkUgo

Length: 36min 53sec (2213 seconds)

Published: Mon Sep 25 2023