Andy Rubin and the Cassie robot | TC Sessions Robotics 2018

The creator of the Android mobile operating system and former head of Google’s robotics division, Andy Rubin wants Playground Global to invest in all things robotics including a demo with the Cassie robot.

[Music] [Music] [Music] So you get a giant robot you bring a Giant robot into your office and Obviously the first thing you want to do Is introduce it to your dog Well it’s my dog so it’s your dog it’s Your robot naturally he has been exposed To robots before and that’s probably one Of the reasons he had that reaction and I think it was actually this was filmed In the the playground parking lot and I Think it was Tim Draper was over to Visit and he was actually the cameraman On that one with the feisty Terrier Trying to herd as a robot dog so that’s Playground global that’s what you’re up To now I wanted to just sort of take it Back a second and discuss your previous Robotics ventures obviously the first Time you got involved with the spot mini And basa dynamics is when you were still At Google acquired a whole number of Robotics companies was Google just not The right home for you know 15 robotics Companies well it was definitely the Right home to acquire 11 I think it was Nine of them public but what if the Total was 11 so obviously what we’re Trying to do is is accelerate the the Effort and doing acquisitions is a great Way to do that once you acquire 11 Companies of any kind robotic companies

Or any other kind the hurting those Things all into kind of a common Strategy a common goal Getting all the companies that used to Have CEOs you know running them to agree On on a product roadmap and things like That are pretty hard I took about a year To get that straight and then it was a I Think well established and I could move On to my next thing what was the common Strategy at the time I mean we’re Talking about again Boston Dynamics We’re talking about a number of these Pick-and-place robots from the outside At least it looked like it was kind of All over the place I appreciate listen you’re a genius I’m Sure you knew what you were doing its Threading the needle a little bit but I Don’t want to give away any proprietary Information there’s a bunch of Confidence nobody’s listening it’s fine Just whisper it in my ear a lot of that Stuff is actually still going on so I Think it’s I’m pretty pleased with what I hear but in general I think with Automation a lot of the opportunity with Automation whether it’s a self-driving Car or robot or you know an industrial Robot in a factory it really starts Making a lot of these problems in Logistics and routing and decision Making it’s obviously becomes a software Problem once you have the actuation if

You think about the life of an employee Who works in an Amazon warehouse all Right what they do is they’re sitting There and they wait for their cell phone To vibrate right and they look at the Screen and it says go to this rack this Level this bin and pick the package and Bring it to station five yeah right so The human goes and does that and then Completes the task and waits for his Cell phone to vibrate they’ve got a Countdown going the machine knows Exactly how long it takes it and they Know how many steps it takes and Everything else so so you know it’s an Efficiency problem and the humans in the Loop and the humans probably the most Inefficient thing right so of course Amazon’s done a great job at automation They continue to so there’s a lot of Companies chasing that but to be the Person in that loop you’re the fault you Know if there’s a MTBF and there’s a Fault tolerance you’re that you know the Spotlights on you and it’s already been Defined as a software problem because From an All the way through to that person Cellphone ringing right is software Right so you know some some algorithm Pick that person because they had Knowledge of that area the warehouse and They’re in the right zone to go and do That task so is it going to be necessary

I’m in order to really advance robotics To a point where they become truly Ubiquitous to have a large company like You mentioned Amazon or Google at the Helm well I think you know initially a Lot of this stuff you know we’ve had I Think you know 1961 the you know motion Unum 8 was the was one of the first Industrial robots right so we’ve we’ve Gone through these various kind of Cycles of innovation and and now we’re In a cycle where there’s some disruptive Technologies machine learning and Artificial intelligence have been Applied to these things and in my view It just so happens these large companies Or you know the mega corporations They’re the ones that are leading the AI Charge it’s a data business a lot of Data needs to be accumulated and Processed and these companies already Have infrastructure to do that the Number of compute power you know the Amount of CPUs and the amount of compute Power in a data center at an Amazon with AWS or a Google whose you know job it is To you know satisfy search requests There’s a lot of compute there that has Been slowly repurposed to machine Learning and artificial intelligence so As somebody who’s back in the the Robotics funding game once again what Are some of the lessons you learn from That first time out when it comes to

Choosing a successful robotic startup Well so you know part of playgrounds Thesis is we’re a nine hundred million Dollar venture fund it’s it’s pretty Diverse will invest in early stage later Stage and will invest in some pretty Longshot crazy stuff and then will will Also invest in near-term stuff and Obviously from a finance world that’s Just arbitrage you want to make sure You’re kind of covering all the sectors With with your capital you know the Learnings that I had kind of coming into Playground for this style of investment Is This isn’t gonna happen overnight right These are long-term bets you have to Take and you know kind of piece by piece It’s like you know building a puzzle Rather than doing the thing that’s the 10 or 20 or out thing there are Waypoints to get to that milestone and You have to invest relatively Strategically to kind of fill in those Puzzle pieces for the waypoints but each One of those waypoints can be a Successful standalone entity a company And that’s what I think venture does Best so so is it looking towards short And long term with each of these Companies specifically I mean again you Know we had mark ravers on stage earlier And they mentioned that you know for Really for the first time in their

25-year history they’re starting to Really look short-term as well they’re Going to be commercializing this robot And obviously that’s what affords them The ability to do these moon shots at The same time as you’re looking at their What about a dozen robotics companies in Your portfolio yeah I think I think I Don’t keep track of all of them some of Them are confidential but it’s about 11 Or 12 right now so do you look at Companies that both have short and long Term potential well look I mean 25 years Ago when Boston Boston Dynamics was was Founded out of MIT or wherever if if you Were to tell them 25 years ago in two Years I want you to come up with a you Know commercial yeah anything they Wouldn’t been able to do it so what’s Happening today is founded on the Platform of 25 years of innovation right So some of the investments we’re gonna Do our investments around innovation Where we think the industry will end up Right so it’s a little bit wargaming out Where you think things will end up for Example you know we have investments in Quantum computing right now you know It’s not feasible at the scale that They’re they’ve decided to tackle to do That today but if you believe in physics You can understand that someday there Will be quantum computers right so Everything else can kind of fall in line

With that and obviously supercharged Machine learning and artificial Intelligence and everything that comes Along with so we can’t just create a Wormhole and have that being profitable Overnight maybe maybe not gonna happen Overnight Certainly you’d want to protect yourself From getting sucked into that worm You see they simply be well positioned Though where you are now to really sort Of you know invest in these moon shots You know or is it important though that These companies do become profitable Quickly in order for them to remain in Your portfolio of course I mean this you Know for me this is graduating from a Company like Google that have the Resources to think really long term into A role as a venture capitalist you know My job is to make returns on the fund Right and I want them to be you know 10x Returns which is crazy by the way so so Obviously we need to be making sound Wise good investments that have that Ever return for our limited partners so You talked about quantum computing and You know we talked about the ACM bipedal Humanoid robots those are a bit further Out but what are you especially in your Portfolio in water what do you see as Being the most profitable short-term Robotics investments well look I mean I Think some of the technologies necessary

To build self-driving cars for example Our robotic technologies navigation Software navigation and new ways to Think about navigation software are Important are important parts of that I think regulatory wise self-driving Cars are you know a ways off I think you Know we’ve all read the stories and We’ve all witnessed some of some of the Learnings in that field but applying That exact same technology for example In a warehouse right it’s not regulated And it’s a lot safer and there’s it’s a Relatively structured environment so we Have a company called campus technology That has written a ground up a Navigation stack just like a navigation Stack for a self-driving car that runs For vehicles in a warehouse ok so you’ve Got you’ve got warehouse logistics You’ve got a drone company you’ve got a Farming company you’ve got agility Robotics that make the incredibly cool Kassie robot that we’ll be seeing later Today is there a cohesive vision across All of these in the same way that you Had when you were at Google well I think You know so first of all I’m a venture Capitalist I’m not a c-corporation so I Do have to think of these as Stand-alone investments so they’re the Cohesive strategy is that we think Automation and intelligence is a Disruptive force AI is eating software

Machine learning and data are driving These things and if you think about the Definition of a robot sense plan and act You realize that you’re in a data Business and that data is probably going To come from the real world rather than From something that’s generated in the Cloud and what do you know robotics and Automation live in the real world and That’s what their job is is to sense and Act where the planning is going to Probably be driven by some neural Network in the cloud so you know that That architecture that technical Architecture is something that is Obvious to us and and our investments Kind of plug into one of those pieces in That technical architecture I keep hearing about how everything’s Accelerating how we’re gonna get to a Point where you’re you know robotics are Going to be more ubiquitous than the Smartphone but one of the closest piece Of technologies as you said is a Self-driving car and you know because of A lot of what’s happening because of These stories that are in the news it Still feels a ways out I mean how long Are you actually projecting when you Talk about this world where robots are Truly ubiquitous well if you if you look At what people are doing with drones Today there’s you know business Opportunities Gallo is one of the

Companies that I think is is here You know there’s opportunity to apply The same technologies the navigation the Obstacle avoidance the following and Things like that you know to consumer Products today if you think about Machine learning and and-and-and-and-and Data and how to apply that into a Consumers life I think it might be a Little early but look you look at even Like a Roomba your robot vacuum cleaner Each one of those internally has a map Of your house so how many people in the Audience have a Roomba right so you know This might not be representative your Lowly I always thought this maybe people Don’t want to admit Now you know you’ve given away a map to Your house but I think some of these Ideas have kind of been germinated in Consumer products and will have a bigger Role in in this disruptive force when The data for these things gets federated In a way where they can serve a better a Better good I keep here to talk about Robotics as a platform one of the things That Android did remarkably well was Create a cohesive platform across Devices create a framework create Consistency among you know any number of Third-party products do we need a model Like that for robotics to continue to be Successful well look I mean I do look I Do look at these kind of waves of

Innovation that become platforms I think software platforms and operating System platforms and things that power Third-party developers creates a Virtuous cycle these things then in turn Obviously create ecosystems that can be Monetized in a variety of ways the Business model of platform ization is Something that is near and dear to my Heart for robotics and automation the Likelihood of there being one cohesive Platform that everybody ends up adopting I’m not sure but if if a lot of these Companies can start agreeing on data Sharing formats for machine learning Then the learnings of these things can Be shared and obviously you know the Transference of knowledge from one to The other happens instantaneously at the Speed of light and in general I think That’s a really really good thing we had Some folks from our OS on before I mean Obviously that’s as close as we’ve got To a ubiquitous robotics operating System I mean is that the way the future Does it make sense to have that Operating system across different Devices I think again if you think about The architecture and how this is gonna Play out it’s gonna be the large Companies that own a part of the AI Cloud right and and these machines Whether it’s your cell phone or whether It’s a industrial but in a factory or

Whether it’s a consumer robot in your House these machines are going to Basically be the point of Data creation if you think of these as Sensors and that data is gonna kind of Flow up to the cloud and be processed in An intelligent way that’s the plan part Of sense plan and act and then they’re Gonna come back down to these things and They’re gonna be actuated the Architecture of this is a lot different Than like you and I are connected you Have your brain I have my brain I have My eyes you have your eyes we go and Have our independent experiences and Then we talk about them right and you And I talking about them and talking About them into microphones to audiences Is like this lowest most painful way to Communicate I’m enjoying it but our Knowledge I’m a little nervous but if That could happen at the speed of light Where there’s a million sensors sensing The world in real-time and they’re all Going to kind of a single brain in the Cloud and that knowledge can be Transferred as many times as we like Well shoot that’s like you know teaching My two-year-old not to spill milk right It took a long time and then when I had To do that again with my second son I Had to go through that same experience Of you know teaching him not to spill Milk but you know in in this

Architecture that I described you learn Not to spill a robot learns not to spill Milk once and then no other robot ever Does it again so what role is playground Global able to play in creating that Cohesive structure so I think you know One of the things we think a lot about So playground is a typical venture Capital firm mashed up with like an Engineering firm so we employ a lot of Engineers a lot of brilliant people Technologists that have invented some of Those stuff that we’re talking about Today and you know their job is to help The portfolio company help the Investments that we make think about the Impact of the stuff that they’re Creating whether it’s the social impact Or a technical architecture or even kind Of pairing them up because they’re Solving adjacent problems where there’s An underlying technology that might be Useful for both of them and I talked a Little bit about federating data how can These companies that we’ve invested in Share data amongst themselves to Accelerate their plan and give obviously Playground as a venture firm And advantage so here we have a special Guest yeah we have a walking talking Guest so one of the one of the companies That we’ve invested in is a company Called agility robotics and I’m a big Fan of biped robots and the this

Technology leverages this notion of Passive dynamics which is going to Become really really important as we Move forward here because passive Dynamics are similar to the way humans In biology and animals operate you know My muscles in my leg don’t look like a Servo motor with harmonic drive which is What a lot of robots use these days There’s a different mechanical Architecture to the way human the human Body and biology works and this team has Done a really really good job of Implementing that in hardware that’s a Mimicking biology in a way where I think There’s going to be a lot of as a Platform there’s gonna be a lot of Innovation built on top of it and this Is called Cassie is the name of is the Name of this robot so I have to ask you Real quick we don’t have a lot of time Left but are you able to look at a robot Like this and see commercial Applications is this is short is this a Robot that has a short-term payoff well Remember I do look at so so in order to Do the things that this robot is doing In front of the audience here there’s Obviously a lot of engineering and a lot Of compute power and a lot of software And a lot of innovation that had to go To do this and like I said before I Think in the terms of platforms right so Why wouldn’t I want to add legs to

Everything in the world you’ve all seen Here these these is it getting more Close to you OEE beeps I there’s a mating call you’ve All seen these delivery vehicles these Kind of last mile delivery vehicles if You want one of those to walk up the Front steps in your house and leave a Package at the door the only way you’re Going to do that is with legs the world Is built for beings with you let I’m insane you know the majority the Population lives in high-rises in these Big urban areas right so you have to be Able to ride an elevator to a front door And that’s just not something that we’re Capable of doing today with wheeled Vehicles so next time you order Something from Amazon this guy might be One hood you may need to build some arms Onto him but you’re halfway there great Andy thank you so much [Applause]

You May Also Like

About the Author: admin

Leave a Reply

Your email address will not be published. Required fields are marked *