Inside a Humanoid Robot Lab | Figure

we see a future where there might be 10 billion humanoids on the planet our mission really here is to expand human capabilities if we have humanoids to the market able to do any physical labor think what that does for the world labor is a choice they'll free you up to do more things you love every day we think that's really an exciting inspiring [Music] future yeah I actually grew up on a third generation farm corn and soybeans in the midwest in Central Illinois I would say for a long time that I really wanted to build things and uh and make something to this world that doesn't exist in high school I kind of it became really apparent that there was like kind of three areas I really wanted to work in AI systems Robotics and internet and so I've been spending the last 20 years building companies trying to help improve humanity and make exciting future I mean we're basically at this Robotics and AI Revolution nobody's been working on this hardest problem in robotics which is like how do we do physical labor what is needed for that is something like very mobile dextrous and be able to build like a basically a AI data engine so that the robot can learn how to do basically more things every single day you know here we've kind of embraced that complexity both in the mechanical and software side and built a very complex platform that can do basically anything physically a human can do I think the conventional wisdom here is when we walk into a customer say like a big Warehouse or manufacturing retail company that we' basically been taking people's jobs and that that couldn't be further from the truth when we walk into these facilities there's a giant labor crisis going on these companies are losing 50 to 150% of employees annually they can't find anybody to do these jobs they just don't know a solution to help automate through these problems we feel a figure the demand for what we're doing is almost unbounded I would say we're trying to be one of the first in the world to deploy us at real scale commercially yeah welcome to figure uh we're building cuboid robots let me show you around a little bit so this is our robot test area uh we have two robots currently live that we're operating every single day our goal here is to mure the robots to a point where they're doing commercial tasks every single day in our facility this is actually the Torso for the first generation robot this houses the computers power distribution a whole host of other electronics so basically we have like a manipulation area here where we're working on developing our current Next Generation hands like ectors on the robot that can basically grab anything a human can grab and do work with it these are all the lineup of our next Generation actual that we designed all in house our largest here is for the knee and hip we make all of this fully in house we've been running them now for the last like 60 days this is the largest motor and actuator on our next Generation robot so this is our fabrication facility so we're able to basically make at this point now any single part on the robot and this is really critical for Speed so every morning we shake out the robot which means it's doing about three or four different types of uh movements to make sure that the latest code is running well on the robot there's no hardware issues as we approach the day and it overall just keeps us disciplined like we need to be operating robots every day this robot here is our first version robot we have five of them in our facility it's a full electromechanical system so means batteries and Electric Motors it has about 30° of Freedom multiple different types of cameras for perception systems right now what we're trying to do is doing fully in toin autonomous applications with this robot so can we do like humanlike work at human speeds over the last 30 days we've be able to demonstrate uh that work here in O so we hit a really important Milestone the Summer where we walked uh a dynamic bipo humanoid which we did in under a year and we believe is probably the fastest to have done in history since then we've done full in to-end application autonomously with our robot it's just getting all the different subsystems across like 10 different teams to all come together to work having to do that on the timeline we did means like integrating a system up really fast and to be honest the last 6 months have been like bring up hell everything that could go wrong has gone wrong we've had suppliers Deliver Us product that was just not working we've had to build 911 teams internally to go build new uh parts of the robot to make sure it works the robot's performing really well now but that's like that's the state of Hardware you got to kind of you know dive deep and fix these problems we really want to build this General interface a robot that can basically interact with like many different types of applications and tasks we want to be able to do the same thing humans can sit in a chair operate Machinery use tools we're really focused here on delivering a really high quality Hardware platform and then basically building a high level AI system on board that can continuously improve the systems and the robot learning over time we very much have a machine learning first strategy around perception systems manipulation policies and overall getting the robot to do these applications that's puts a lot of pressure on us to build like basically the right AI data engine from scratch and build the right training sets for the robot to make sure it can scale successfully by itself in any environment uh figure we're a year and a half old we're about 70 employees one of our core floss phases moving really fast the hardest part we've had last 18 months has been there's not really a lot of commercial supply chain for humanoids so what it's forced us to is basically design almost entire system from scratch outside of maybe the compute everything else besides that we we designed here a figure and that's uh as you probably well aware a very steep learning curve so we've had to design the actuators basically from scratch the Battery Systems operating system control software a lot of electronics are designed from scratch basically designing what is the best robot possible we can get get out the door I think I have like two big philosophies that are really important to building out this area of kind of De Tech now the first is you need a really great team so I spent a lot of time recruiting what I think is the best engineering team in the world and two is I spend all my time on product and Engineering at the end of the day that's like the most important thing we do is ship a product that's useful or service is useful to in customers we have like a really unique culture here where we want to move fast we want to be super optimistic we want to do like really great engineering work the culture we had instilled here is just not for everybody we're in the office basically 5 to 7 days a week we work really hard we really care about working on the product so we're really looking for that unique person that kind of wants to go out and put a D in the world we've now assembled about 70 engineers in the team that I would say are the Best in Class of what they do what's going on over here yeah uh teams does push-ups twice a day why I I don't know we have this philosophy here that we talk about is like the only way out is through I think for entrepreneurs if everybody knew how hard it was they won it's it'd be unclear how many people would actually go out and do this and some way like that that level of like uh ignorance actually kind of helpful my advice is just get out there and start building start building stuff that subscale start building that stuff in simulation start getting just something built start learning it's like a stair step approach over time right like you're learning you're recursively getting better you're making less mistakes you're progressing the technology you're raising more Capital you're getting better customer feedback so I think my biggest advice is to get out and start building as fast as possible the problem we have here is a humanoid has never been commercially viable never been a humanoid put into a real application getting paid real money to do work the core focus of what we're trying to do here at figure is be able to produce a commercially viable robot into Market we think in the next 12 months so that means getting a robot fully autonomous system set up and integrated into our clients facilities doing in toin work hopefully up to 20 hours a day without any human assistance so we spend a lot of time at the requirements level trying to understand what our client needs are if we're going to ship robots into warehouses manufacturing different levels of retail we have to know like what actual work we're going to be doing that basically helps Define what the robot needs to do from the performance safety cost manufacturing so we'll do different trade studies on should we use battery cells what type of form factors should it be cylindrical prosmatic pouch should the motors be linear or rotary there's just a whole host of different decisions that we have to go and make I would say for the first alpha build that we did we've made about 70 of those decisions last year like what what we should do we then went out and did a bunch more testing we built the robot and we're slowly validating all those decisions to make sure we're right all the right decisions are being moved into the Next Generation robot and all the wrong decisions were being relooked at and re thought out and put into the next robot as well we have a really strong roadmap to support a very fast-paced commercialization plan we're all working 5 to seven days a week now trying to go hit that we have to get a robot into our client facilities doing real actual human-like work needs to be safe it needs to be reliable we have to do that an environment that's really like high variability really dirty everything that could go wrong will go wrong getting the robot to work in those type of environments is going to be extremely difficult for us to do and that's the next big hill for us to find I think of things is like what what kind of impact can I have when I'm alive what has like the highest lever arm for that impact technology has the ability to touch like billions of people's lives at this point so my advice would be to kind of find those areas that are kind of new and novel spend a lot of time like really deep in the rabbit hole trying to figure out like what that looks like get your hands dirty more answers will come the more time you spend in those environments make sure that you're really in it for the long haul like this is really a 10 plus year Journey this is not a fewe thing so make sure you see yourselves in 10 or 15 years in that industry doing this thing and being happy I think it's going to be really important uh so at least for myself like I think this is probably a 20-year Endeavor 30y year Endeavor my rest of my life I couldn't think of anything more exciting and more challenging to work on and that gets me really excited to wake up every day the next couple decades you're going to see humanoid robots basically going from kind of the science fiction view to like in actual real life as we approach the next 10 or 20 years every person in the world has a humanoid and that humanoid will do everything for you physically uh grabb you the coffee go do an errand whatever you need to do it'll just be like having a phone humanoids here have the potential to be the largest market in the planet technology if not pushed up the hill doesn't happen so somebody's got to get out there and really push the envelope and it takes a lot of effort from a great team and we've assembled here probably one of the world's best teams to go after this very hard challenge looking back when I'm like 80 I want to look back and say where did I spend that time what kind of impact I made and has it been helpful and I think at least today it feels like these are the right projects for me in this time to be working on to make that [Applause] impact that was insane I've been wanting to feature figure for a long time I have this like secret list of like top 10 companies that I want to film and figures very high on that list it's weird how fast the figure team has been able to move in less than a year going from like a concept to a robot that's walking and then saying in a year we want to be like in customer facilities working and learning through that process this is such a perfect episode to End season two 20 episodes week after week uploads from S3 I think this just embodies I think the peak of what it means to build something really exciting in the heart Against All Odds Against All common wisdom these are the exact types of teams at all sizes we want to fature on S3 and so having this be a company that that we End season 2 on is is incredibly meaningful to me it's also just insane that we made it here to the end of season 2 we've been doing this for 20 weeks it flew by I feel like I blinked from episode 5 on and now we're we're here I don't quite know how we did this I do know that this would not be possible mentally for me if I didn't have the support encouragement and interest of everyone who's watching this and who's a fan of the show and a fan of this Tech space that we're trying to improve the storytelling on so from the bottom of my heart thank you it's incredibly motivating when people reply like argue about the startup and the in the replies and just in general can at least agree that this is what the text based needs that's really my mission right now is like how can we provide better storytelling for groundbreaking service teams with us3 and then for for the news side of things how can we like better capture the amazing stuff that's going on in this space and get the rest of the world as excited about it as we are from the bottom of my heart thank you for being a fan of this thank you for watching thank you for supporting and please keep doing so I'm really excited to keep growing this and see where we are in another 20 weeks so yeah 20 episodes end of season 2 beginning of season 3 starts next week thank you all for the support and let's keep building the future

Channel: S³

Views: 54,302

Keywords:

Id: 48qL8Jt39Vs

Length: 11min 27sec (687 seconds)

Published: Sun Dec 03 2023