#070 – Fully Autonomous Drones with Adam Bry, CEO of Skydio



join attendees from Shell Exxon Mobil Chevron Duke Energy and 800 more energy UAV leaders at the second annual energy drone coalition summit in Houston Texas on June 19th to the 21st visit Energy drone coalition dot-com to learn more and get your ticket welcome to commercial drones fm the podcast that explores the commercial drone industry the people who power it and the concepts that drive it I'm your host Ian Smith hey everybody Ian Smith here and I am sitting in Redwood City at the HQ of sky do with Adam Bri the CEO and co-founder of the company and welcome to the show Adam thanks you and thanks for having me on thanks for having me at your office it's really cool spot very aviation themed kind of like a cool warehouse theme to it as well let's start off with your background so tell us a little bit about yourself about Adam and your background and how you really kind of got into or chose drone technology and became the CEO of a company called Sky dia yeah so I've actually been building stuff that flies since I was a little kid so I grew up flying radio controlled airplanes and I took it probably way too seriously I actually traveled all around the country competing in aerobatic competitions I did that with my dad so the balsa wood with the heat shrink and the heat yeah yeah the whole thing so I spend a disproportionate amount of time in my youth in the basement sort of slaving over these things putting them together and then and then flying them as well so I certainly didn't start doing it because I thought it was it was gonna be a career I mean I just love stuff that flew I love the technical challenge of putting them together and then I also really grew to appreciate and love the challenge of learning how to control them and fly them and that's kind of what got me interested in engineering and then I was fortunate to just be going through engineering school around the time when it was first possible to take computers and sensors and put them essentially on RC airplanes and and start to write software to make them do things autonomously and just was totally captivating and addicting to me so I was a grad student at MIT starting in 2009 and really the goal of my master's project there was to do something that would be impossible for a human pilot to do to build an autonomous system that could do things beyond the capability of the best human pilots so the main output of that project was an airplane that could fly itself through a parking garage saw the insanely cool video about that how long ago was that video released so that video was 2012 Wow I yeah I was 2012 it was it was about a two and a half year project I mean we yeah there were a lot of different aspects of the research and it sort of worked a year and a half in but it really worked two and a half years in which is one we made the video that got the most most sort of play on on YouTube yeah and if anyone hasn't seen that it's basically like I'll take the liberty here to describe it from a layman's perspective but it's a fixed-wing drone flying and I cap act parking garage it seems and literally flying around inside of like I don't know how tall this the the ceiling was like 10 8 10 feet 12 psi I know it in meters because our lab worked and after it units and we had detailed maps of the whole thing so I think was about three meters I was two and a half meters floor to ceiling and the airplane had a two-metre wingspan so when it was banked over oftentimes you can see this in the video like it was you know 20 30 centimeters clearance for all these obstacles flying at about 10 meters per second about 25 miles an hour so this is this is something that a human pilot like maybe could do laps maybe once or twice if you're like really talented right I think I think not even I mean I so I was a I was a pretty good airplane pilot in my day at least when I was competing and really the sketchiest part of the whole thing was when I was trying to fly this thing so like sometimes to collect data I would have to fly it around manually and I crashed a lot doing that so the computer crashed too but I found it I found it basically impossible to do it flying it flying it manually that's awesome and so ok so after this then you decided there's there's a business here maybe so actually not not directly to sky do so you know at the time at MIT and this is where I met my co-founders at the time at MIT we just thought this stuff was super interesting and fun and challenging and you know generally it seemed like autonomous flying stuff had a lot of potential so it was exciting to work on but I was a I would say it was before most people had really seen a lot of the the business use cases that seem exciting today and after MIT we got the opportunity to go to Google and help start project wing there which was Google's drone delivery effort so a B'nai abe's our CTO he's one of my co-founders we were at Google for a year and a half working on on that project we left before it went public before they did the first public demo to start sky do nice and so tell us about a little bit about sky dia so how old is the company now how many employees you guys at and obviously we didn't really mention yet but you guys have released your first product the r-1 which is a fully autonomous flying quadcopter that's super badass but you had tell us a little bit about sky do first and you know just the company in general and employees and such yeah so the basic idea for the company was pretty simple like we you know we looked out at all the exciting things that people wanted to do with drones so drone delivery was one that was getting a lot of attention obviously we're working on it but 2014 was starting you know when people are getting excited about different kinds of commercial applications the consumer use cases we're starting to come into focus but basically what we realized is the things that seemed most exciting to people were also going to be the ones that didn't work particularly well with manually flown drones and it seemed like autonomy like really good trustworthy autonomy was just gonna be at the core of most of the most interesting valuable exciting drone applications and it was something that wasn't really possible with existing products and I think it was possible in ways that people didn't totally understand like because you know no drone has a person in it there's not this sort of there's not the same direct connection of like a self-driving car versus a manually driven car like people look at it and it's easy to imagine it doing all these things even though really doing them might take deep onboard intelligence in a way that that at the time certainly didn't exist so you know autonomy autonomous flight was something we knew a lot about something we loved working on and it just seemed like a great opportunity to to take a shot at building a system that could be a big enabler for the industry and that's really what we've been focused on since we started four years ago so you know for the first almost year it was just three co-founders in a basement writing software an off-the-shelf hardware the company's grown quite a bit since then so we're now about sixty people mostly engineering although we do all you know building and shipping a product is a significant undertaking and we cover all the bases that it takes to do that now it's actually I think a very small team for the ambition of the product that we have you get it's building it here at the headquarters okay yeah yes we actually do final assembly and test ourselves which enables us to ensure super high quality and it's also enabled us I think to just ship faster than we otherwise would have because we can just we can iterate so quickly like if there's an issue on the line the software engineer who's working on that component can go and inspect it diagnose it debug it and fix it in a way that you know just takes a lot longer if you're if you're trying to do it overseas yeah true that very cool and so the specs on the are one so it is a quadcopter in the traditional sense yeah it's got 360-degree obstacle avoidance 4k camera autonomous flight built-in hard drive for storing media and such what else is going on like under the hood you know what there's there's like a bunch of cameras on this thing I mean tell us a little bit about that so I think the starting point is there's kind of a philosophical difference in how it was designed and built from other products that are out there and basically what I would say is that we think of it more as a flying computer then as an RC quadcopter so I think most of what you can buy today still kind of has its in this radio-controlled quadcopter paradigm really which comes from RC helicopters and RC airplanes which I think has some obvious and some subtle implications for the hardware and the software so on our side with r1 you know thinking of it as a flying computer and something that you want to be fully autonomous that means that you need omnidirectional sensing like if you want to be able to trust it to fly itself and you want it to be able to fly in any direction it needs to be able to see in every direction I think that's just sort of like an obvious starting point but it's not something you know obvious to us but I think it's not something that like exists out there otherwise today and then you need a really powerful central computer you know those are sort of the backbones of all the computing devices that we use and benefit from on a daily basis so the r1 is built around an Nvidia tx1 which is a super powerful CPU GPU combination with the Jetson yeah okay yeah it's part of the Jetson family and that enables us to leverage you know all the cutting-edge frameworks in deep learning and computer vision that are that are out there and and really you know build a super powerful flying autonomous system so what about the R&D process then for the r1 you know how many years in the making was this before you guys actually you know released it what kind of challenges did you come across I mean like lessons learned I'm sure there's so many of them more than you can even list but yeah can you give us a little bit of color behind that you know three people three co-founders in a garage kind of phase where you were just testing and iterating and it probably a super quick rate just 24 hours a day yeah so you know overall I would say the it's it's taken us I think unsurprisingly for like a hardware startup it's taken us a little bit longer than we thought to launch the first product but the other thing I would say is that the capabilities and the level of robustness and maturity and the autonomy is really beyond what we thought it was gonna be in the first product and I think that were you know it's a testament first to the team that built it we're fortunate to just have an incredible group of people here spanning software and hardware and all the different disciplines it takes to put these things together but I think that we're all so fortunate to be just doing it at the right time where we're riding these trends in computer vision and deep learning and the hardware to support that stuff and I also think that there's this this relatively new phenomenon where historically most cutting-edge robots are built in academic labs where you have a couple of grad students usually and I know because I used to be one of them where you're sort of trying to manage this off-the-shelf hardware and hack it to get it to do what you want but really you want to be focusing on the flight software and the algorithms but in order to demonstrate the robot thing you're doing you have to design and build the whole thing and if it crashes you spend a week fixing it and it's really like pretty slow going and at sky do we're fortunate to just have a world-class team of people across hardware and software and embedded infrastructure and like all the things that it takes to build one of these systems and on the autonomy side we have literally like world experts in 3d geometry computer vision deep learning motion planning control and each of them is working towards the common goal of what the system is capable of and the maturity and sophistication of each of those systems is sort of reinforces all the other ones as well like when you have when you're doing really good motion planning and really good computer vision side-by-side they both benefit each other and so I think that all of those ingredients together has been has been pretty fun to see what what you can do and I think that it's accelerating it's not slowing down like you know I've you know I have the long view on this because I've been around since the beginning when it was just three co-founders in the basement like the pace at which we're getting data back our level of understanding of the physical challenges and our ability as a team to like build stuff quickly that improves capabilities and adds new features and stuff is is getting better and faster and so I think it's like it's just a really exciting thing to get to be a part of the other thing I would say and you kind of alluded to this you know when it's three co-founders iterating on this stuff you know one of the sort of methodologies that I think we've adopt did or I would say adapted to what I perceived to be the challenges of building a system like this there's this idea of like the Lean Startup methodology which you know people take in a lot of different directions but it's basically like you want to ship your MVP as passes as fast as possible and learn from customers and I think there's a lot of you know there's a lot of benefit to that way of thinking but it doesn't necessarily apply to all systems and I think in in robotic systems one of the sort of adaptations we've made is that you sort of have to think of the real world as a customer and so we've made a lot of because you know you're building a system that interacts with the real world and has to deal with real physics and perceive things and so you know one of the things that we've sort of embedded in our development process is you want to be iterative in that respect of like what's the simplest algorithm we could use to accomplish whatever we're trying to accomplish get out there and fly a test it look at the data see what's working and what's not and I sort of look at is you you have these two two masters that you have to please like reality and physics and then the actual customer and end user of the product yeah and you have to have like an iterative mindset with respect to both well your approach is fascinating this is really cool perspective that I don't think we've had on on the podcast before and I'm really excited to dig into some of the potential commercial applications before we get to that part of it you know on how to apply your technology commercially there's a minor detail that I noticed I'm not sure how many others notice this for some reason I guess it's just my aviation background with rotary aircraft but I noticed you guys have a six bladed rotor design so each each rotor and each motor is attached to a six bladed rotor as opposed to like the traditional what you see on these quad copters these days you know like industrials grade etc that it uh they usually have a two bladed rotor design like a like a standard helicopter you know in most cases what's the what's the reasoning behind this is there am I looking too much into this or did you guys do testing that was like okay this is aerodynamically more efficient maybe we'll get more battery life or it's quieter what's what's going on there yeah so it was it was definitely very intentional engineering trade-off and I think that it just has to do with the specifics of our form factor so we have these perimeter carbon-fiber blade guard structures that go all the way around the device which hold the stereo cameras that see in every direction and so having those puts the cameras in this totally unobstructed view where you can see everything in every direction and then they also protect the propellers but in doing that they they put a little bit of pressure on the sort of diameter that you can have and so it puts more design pressure on like how much lift can we cram into a smaller area and if you were willing to spin the thing superfast and make a bunch of noise you would be slightly more efficient with a two bladed propeller but the more blades basically like the more surface area you have in there the slower you can spin so the quieter it's gonna be hmm so we did a lot of Investigation and we actually studied it from a few turf angles there's the just the pure metric efficiency and then there's also the more sort of subjective not just the magnitude of the noise but it's people the way people perceive it and so we did a fair amount of sound studies around that and picked a propeller design that was almost as efficient as the optimal thing would be but significantly quieter and and more pleasing to listen to so I just have to say I'm very proud of that question that was a good question yes came up with but no honestly I like drones they're noise is it's very disturbing and it's one of the one of the hurdles to like mass adoption like I mean you know you can't just imagine there was a study like NASA did a study someone someone did a study that like the noise of a drone is like more annoying like than anything or something they compared it to something super mundane like the sound of just like a car going down the road at the same same frequency I don't know if I'm using the term correctly but basically the drone sound was like grating to the ears and kind of annoying and so yes I'm looking I'm very glad you guys thought so much into that and did sound testing too because I think that's a big hurdle as we move forward as an industry in the future yeah I totally agree and I think I mean I think we'll come back to this but I think that you know it affects size weight noise all come together and that affects our sort of view of the commercial landscape as well cool so the price of this drone that's something that I've you know I look at the internet comments and everyone's like oh it's so cool but oh my god it's so expensive they're never gonna sell any of these things it's probably purposeful if I'm not mistaken I mean you guys price this like knowingly like is there is there like a strategy right now like do you want to limit the demand a little bit with this you know it's your first hardware product it's the initial thing it's kind of nice you know it's it's very powerful what's the story behind the pricing sure so I mean I think the right way to think about this is it's analogous to other kind of first-generation compute devices so there is a lot of technology packed into r1 to make it possible like the goal of it is not to pack in as much technology as possible it's to deliver this break through autonomous flying experience which enables all these cool things for our customers but in order to do that we had to do a ton of R&D and we have there's just a lot of from a hardware standpoint there's a lot of stuff in there so and I think that's you know that's just the reality of doing these like new challenging ambitious products for the first time like if you go back through the history of personal computers I think this applies to Tesla to some extent like they had to build these lower volume more expensive cars in order to figure out how to build higher volume cheaper ones and I think that all of those dynamics are in play for us and the other thing I would say is that it's you know people like to compare it to other drones but I think in a lot of ways that's just kind of a non-sequitur comparison for most use cases because the way that you use it and the kinds of things that it can do are just completely different mm-hmm and and you know the fact that it has I understand where it comes from like it has four propellers and it has cameras on it so it looks like a drone and you know it is in some sense a drone but it's just very different from the other stuff that's out there and I think you know it's it's clearly at a price point where it's not like a mass-market mainstream thing but has a you know we feel and our first customers would agree like it has a pretty amazing value proposition for what it can do where it really is like a professional film crew thing that fits in your backpack that one person can use anywhere in the world anytime you want without someone needing to be there to fly it so to capture activities that people are passionate about to tell a story in a way that otherwise wouldn't have been possible you know it's probably not yet for everybody but we think that it does something pretty special for the people that that you know have a have a use case that fits speaking of the you mentioned that you have to think about the way people use it so the interface with the drone of course is like the mobile app that you guys have there's a software approach to this and what was your what was your feeling here I mean with this very highly autonomous aircraft that can you know fly pretty much anywhere what's the like what's the user experience what's the approach of how I interact with one of these and caveat here I have not flown one yet so I'm hoping before I leave maybe I can you know take one first spin and try to crash it as I see everyone try to do and fail miserably yeah we know like that yeah a little bit about the software approach and how you guys thought about that and what you see moving forward yeah so a couple things about the software I think the starting point for us in thinking through the whole user experience which is most directly reflected in the mobile app is that it should match our expectations of how we interact with our phones normally and other kinds of modern technology devices so in particular for for our one you know we we drew a lot of inspiration from just the camera app on your phone and a major design goal is like if you if you're comfortable using the camera app on your phone you should be comfortable using our one which is only really possible with super advanced autonomy on board like you know in a normal drone flying experience you're basically responsible for everything it looks like a cockpit it's like here's how fast you're going here's like the wind speed here's everything else that around you it's exposing all the information that the pilot needs to make decisions for us a lot more of that is taken more you know taken care of by the onboard software and so you can interact with it at a much higher level of abstraction of like here's where I want you to be here's how I want you to move relative to me here's what I want to look at and then and then it does the rest and then the other thing that I would say is that if you're not gonna have somebody manually flying it or at least that's not the the dominant mode that the products gonna be used in then the it basically becomes a software-defined experience like the wave the drone behaves how it reacts to what's around it how that information gets exposed to the user is all done through software and it's all very flexible depending on how that software is written in and what the application needs are and we think in the medium and long term that's a really exciting thing because there's the potential to just automate a lot of different kinds of things that people might want to do for different kinds of tasks through software so we're gonna get to the I'm really excited to get to the commercial applications part but that brings up one of the questions I wanted to ask you know since the drone is kind of like not exposing all this information or at least the software to the the user quote-unquote I'm not sure what to call them in this case because this is gonna be a regulatory question is this user the pilot of a drone if the drone flies itself who's the pilot is it is it me if like like later on is it sky do sky do responsible if like my drone crashes into something or you know breaks a rule like what's there there's a lot of and I'm sure this is you know it's a really tough question to truly answer because we don't I mean as an industry we don't really quite know completely yet it does have parallels to the autonomous car industry and these kinds of things self-driving cars but what are your thoughts on this how does this how does this work so the short term answer to that is pretty simple and it's basically the product has been designed with all the regulations in mind and the user the person who's holding the phone is the pilot in command and if you need to you can take over manual control you can tell it to stop what it's doing you can move it around you can land it and in fact the way that it it flies and the use case basically ensure that it's always going to be with visual line of sight and it also is not going to be flying high it's unlikely to interfere with man aircraft so we think from a regulatory perspective with our one we're in like a very kind of safe place and actually pose fewer questions and risks and challenges than a lot of the things that are sort of most concerning about you know the public safety risks of the way people sometimes use drones today in particular interaction with manned aircraft you know I think the medium and the long term answer is potentially even better which is that I think autonomy is potentially one of the keys of basically encouraging or enforcing responsible use like if the thing is flying itself it's very easy to ensure that it's gonna fly itself in a safe reliable way you can program the drone to kind of like not break any rules yeah exactly and you know our hope is that this can be one of the things over time that kind of opens up public perception around these things a little bit that drones go from being you know the unreliable thing that my neighbor uses to like spy on me to just these like trusted elegant intelligent flying machines that that always do the right thing and that are useful in in whatever they're doing so there does seem like okay as we move forward and this becomes more and more commercialized and more and more autonomous and your technology matures and so does the rest of the industry do you foresee maybe there's like some certification that you might have to run through from the ik like FAA or something that basically says like okay your system has certified that in this scenario it does not do XYZ run into person's face or car or you know we hit tree or power line or something like do you foresee certification as like kind of a I wouldn't even say it's like a challenge it's gonna be I think it's gonna be a milestone maybe for for a company like you guys yeah I think it's a milestone and I think it's an opportunity for us and really for the industry as a whole and you know I think it probably is likely that things like that will exist at some point it's it's a challenging thing to codify in some cases but you know the initial steps that we've seen the FA take I think have been pretty positive and that basically what I would say and I think that my sense based on what they've done is that they're thinking this way too is that you want it to be based on data and you want it to be based on sort of like quantifiable tests that can be done like the I think the wrong way to go about it would be to try to certify the software a priori based on sort of static checking and things like that because it takes incredibly complex software systems to deliver like trustworthy autonomous flight and I think that you're much better off certifying based on the outputs of that system like how it behaves in different scenarios what it's capable of can it meet certain criteria then trying to certify it from the inside out but I think that they get that and I think there's a pretty close analogy to self-driving cars there as well and so yeah I think that we're probably a few years away from that being a real possibility but I'm optimistic that at some point we'll be there and then it'll be it'll be a good thing for the industry I mean I think ultimately anything that encourages responsible use and gives every day people drone users are not sort of peace of mind that these things are gonna behave intelligently and responsibly is gonna be a good thing for the industry this episode of commercial drones FM is brought to you by FLIR pretty much everyone in the commercial drone world knows about FLIR's thermal imaging technologies and may even have a basic understanding of how their sensors work at commercial drones FM we think this technology is deserving of the spotlight to reveal some of the incredible things thermal imaging drones can accomplish drones equipped with thermal imagers offer a way to see the world from a different perspective which can help identify infrastructure problems locate missing people and increase safety and efficiency across a variety of industries the value of thermal imaging for commercial drone operations is undeniable so I want to use this podcast as a platform to dive deep into the world of heat and seek out the best practices and the unique use cases that make imaging sensors the most drone ready payload for commercial drone applications right next to RGB cameras so join us over the coming months as we invite FLIR and their customers into our studio to discuss the technical and practical uses of their groundbreaking efforts and aerial thermal imagery this episode of commercial drones FM is brought to you by Devron UAS Devron UAS offers enterprise drone services for farmland across North America I recently spoke with bond well one of Devin's largest customers so my name is Jennifer Thompson I work for bond while America's long life we're a vegetable processor bond was a very large company with 54 plants in the world here in North America we have 11 plants doing frozen and canned vegetables so naturally I wanted to know how is devran helping one of the largest vegetable processors in the world by using drone technology so we use devaron to service and support our precision agricultural needs for us it's not easy to have our own iron fly our own drones to gather crop data so devran is ready and available to work for us when needed we fly thousands of acres and managing those acres it just works very easily with them in our fast-paced growing season they're not just providing data they're helping to solve issues and find solutions well that's very important work but what's your favorite crop to eat my favorite crop to eat a sweet corn I will eat speak corn 365 days of the year I'm not sure if I've ever eaten sweet corn yet that was monitored by drone to learn more visit Devron UAS comm / drones fm and see for yourself how devran is leading collaboration in agriculture with drones that's Dever on UAS comm / drones fm ok back to the show so r1 is by all for all intents and purposes at this point in time a consumer product like that's how I view it that's how I believe it's marketed I don't know if you would disagree with that but now let's get to the exciting commercial stuff this is commercial drones fm you know what what kind of applications and what is your view mid to long term on how sky do technology you know you guys could license your your technology I mean there's like a software there's an algorithmic component of this you could build your own hardware you could do software I mean what kind of thoughts and and where do you feel that this super highly autonomous technology that you guys have been building and released can be applied to commercial scenarios that are gonna you know save businesses money or increase productivity what can you tell us about that yes you probably wouldn't be surprised to know that we're incredibly optimistic about this and our you know our core belief as I said is basically that autonomy is one of the keys to making drones useful essentially across every application and you know we view is we view commercial opportunities is a huge part of that and in the long term potentially the the bulk of that and if you look at what's working today in the commercial space most of it is sort of built on this paradigm of it being a low flying satellite where the drone is flying high overhead it's above all the structures around it it's relying on GPS and doing GPS Waypoint following and even so you still need to have a person there to watch it to make sure nothing messes up to manage takeoff and landing and I think that we will look back on this is really the super primitive early days of drone use in the commercial space and autonomy is the thing that you know what what all commercial use essentially revolves around is capturing data that you wouldn't otherwise be able to get in some form either doing it more efficiently or enabling data capture that you you couldn't otherwise do everything except delivery I should say mmm-hmm which is sort of its specialized thing and you know we think autonomy is really the key to making that a scalable reliable trustworthy thing where you can push a button and deploy one drone or a fleet of drones or you don't even push the button it's all automated in software and so that's definitely you know a future that that we're building towards and I think r1 is a significant step in that direction so I mentioned this earlier like it very much is a software-defined experience and the software that were shipping and marketing currently is oriented towards this like autonomous capture thing which is amazing and cool and we're seeing a lot of incredibly exciting stuff happen there it's a product that has kind of been imagined but it hasn't been made real we feel like until until r1 existed but we think that there are a bunch of other software to find experiences for different applications which are also incredibly exciting and cool and I think one of the one of them I just overall most optimistic data points on drones is general is that this still happens today even though I've been doing it for the last four years like I'll talk to somebody and they'll propose an idea for something that like I never thought of before for what you could do with a drone and I think most of those things over time are gonna come real it's just gonna take a sort of foundation of autonomy and then creative software developers around and on top of that to enable them and so that's definitely a future that that we feel like we can be a big part of of enabling I like that to whenever someone comes up with something you totally just never thought of and it's like what why didn't I think about that it's really cool the you so do you guys have any like specific maybe like things like so you mentioned okay right now the drones are kind of like hovering they're like above everything they're above your head they're doing remote sensing from a significantly you know in the grand scheme of how drones can go really close to things a far distance DZ so you know future sky do products maybe it's something that's gonna be like right up next to a I don't know power line pipeline wind turbine something that's just literally using a onboard sensors and computer vision to get super super close to like the most my noot detail and just like dive right in and maybe even do onboard processing so it does like a you know a quick pass over the whole thing to get like a macro view and then it goes micro to inspect like some rust or like a crack that it saw I mean oh is this kind of like in line with what you guys are thinking maybe yeah definitely I mean I think you're you're hitting a lot of this stuff right on the first thing is that you know navigation should just be taken for granted like you should be able to trust that the thing is gonna fly itself in close quarters it's not gonna like fly away if it loses a GPS signal it's not gonna bump into anything like I think that's just a fundamental baseline capability that didn't really exist before r1 that is a necessary building block for a lot of these things to make sense and then on top of that you pair the capability to respond in real-time to what you see and I think that you get to some pretty powerful places because you can you can do the things that you would do if you had like an expert operator or pilot or photographer that they're doing whatever the task is so you know the stuff that you're mentioning and you know you can list off any sort of infrastructure but just getting in close to structures seeing what's actually happening capturing them at high resolution in in a seamless safe reliable way just that capability alone I think is something that doesn't exist in a reliable scalable way and can have a huge impact across across a bunch of different industries and and like honestly most of today's drone commercial drone applications they're all focused on outdoor activities so we're not even touch we haven't even talked about like the potential implications for I don't know warehousing and you know zooming around inside a warehouse QR code scanning and just the gathering data I mean in a GPS denied environment because your your product can work indoors if I'm not mistaken like yeah no GPS yeah so the product has GPS on on it it can use it if it's available but it can fly completely independently so it can totally fly indoors you can find confined spaces and you know I can't say too much in detail about this book like we had some notions that we thought places where this could be useful and since we launched we've gotten a huge amount of inbound interest for different kinds of things it can be done like this and yeah I mean we think it's we think it's pretty cool and something that we're in a kind of unique position to be able to deliver I would say the fundamental capability again excluding delivery but the fundamental capability if you think about it is just sort of automating access to the physical world in real time like being able to put a sensor wherever you want whenever you want in a reliable trustworthy way which at maturity yeah I mean there's there very few limits to to what you can do there yeah and you know everything's so fixated on cameras on cameras let's take pictures of sick videos but like you said it can be any kind of sensor like it doesn't have to be a camera that's one of the things I mean in the past five years I've been in in drones or I don't even know how long it's been now but it's just camera camera camera so really excited you know drone delivery is cool because it's not just like using a camera to do something it's like moving something from point A to point B and yeah I mean I would even say that is like some form of automating access to the physical world like it's basically you can move things around programmatically in reality using software like that's what drones enable so as far as so are you guys planning on potentially opening up some type of SDK or way that maybe I'm a developer listening to this podcast or I'm another company this technology sounds super interesting and I can build on top of it potentially what are your plans as far as a sky do r1 or just any future product as a platform so I think it's almost just a no-brainer because there's so many there's so many different cool things that you could do with it there's just no way that you know us or anyone or anyone company is gonna do a good job building all of them and at the same time there's a fundamental core set of capabilities that are useful to all of them like the ability to fly yourself not run into obstacles be able to like do real time perception on what you see like these things are broadly useful they're useful for filming a football game and they're also useful for inspecting a wind turbine and so I think it just it sort of James out platform yeah kind of dynamics and so easy question yeah so you know this has been part of our art vision from the beginning and I can't say anything specific today other than we think it's a real thing we think it's important the other thing I would say around that though which has also informed our strategy is that we think it's important to have a product that stands on its own even without you know applications or whatever running on top I think most successful platforms have that attribute where you know it might only be realizing some percentage of its potential but it can do useful cool things for some users on its own and that can sort of get the flywheel going of like people use it and play with it in our case like we get feedback and data back and we can improve it and then we can start to open up the capabilities for other people to use and extend in different ways I'm excited for this so the future I mean like obviously that's one of the one of the kind of prongs I guess of sky DEA's future but what else I mean are you excited for just in general you know looking forward you've you know I heard about sky do probably when you guys finally went public with what you hey we're building this really cool autonomous thing stay tuned and like that was like a couple years ago or something I might be messing up the timeline but what are you excited too excited for as just looking forward for you know after having launched the r1 and you know more people people are getting it in their hands they're talking about it yeah I mean the the really exciting thing to me is that I think were we're still you know we're still close to the beginning like we're still close to the beginning of what's physically possible with these products they're still huge Headroom in the software to improve things and make for an even better or smarter more seamless experience there's we haven't I believe even with what's out there today we we haven't really scratched the surface of applications because in order to develop applications on existing product you the application developers still sort of needs to think a pilot and really the application developer should be thinking more like they're writing a videogame or something like they're just moving around this this virtual sensor platform and they don't have to worry about taking care of any of the stuff that a pilot would need to take care of that's all taken care of by the platform itself and so I think that we're just like we're right at the beginning and there's there's just exciting stuff in every direction so you know that's kind of a vague answer but that's that's really how we feel about it and you know it's exciting I think the other thing and this is maybe just a little bit more tech philosophical in general I think that a lot of a lot of technology well technology goes in all different directions but a lot of it is sort of pulling back from the physical world it's like AR and VR and just like getting the screen closer to either face so you don't have to interact with reality and not you know that there's anything wrong with that I think there's a lot of cool stuff happening there but you know my hope is that drones can be at least one avenue of tech that that anchors us more to to doing stuff in reality the drones move around in reality like our first product and with a lot of people inside the company say this like a lot of people here have been a lot more active since they started using our product because it's just like oh like what am I gonna do this weekend I'm gonna go do some like fun outdoor thing that I can like test r1 and get some cool footage and like and you know I hope and I think that that's a that's a real thing that that will continue at least for for some use cases well I'm watching you guys super closely I'm really really excited for what you guys come out with next and how this is gonna affect the commercial industry the commercial side and industrial side of drones so if you're listening and you're interested in extending sky dio's r1 for commercial applications go to sky do.com slash technology and then scroll down to the bottom and then there's a build on our one module and you can check that out leave some feedback over there on the kind of like how you would use r1 for commercial purposes is that correct yeah I mean we're basically just looking to capture our interest from people because you know as I said this is this is something that we're interested in we are going to do and so you know that's sort of the way for people to be first in line to like hear about it and that's a real thing like I read I read what people write there and it you know it affects the way we think about what order we should do things in awesome so that's sky do.com slash technology so Adam Bree CEO co-founder of sky do thank you so much for for having me hosting me here thank you for being on the podcast I'm really excited to go fly that r1 now all right sounds good thanks again alright everyone we're gonna cut off the mics Cheers you

Skydio is a very special drone company with a thought-provoking inaugural product. R1, the company’s first drone, costs $2,499, is fully autonomous, and it’s marketed mainly to hobbyists.

So why does a startup’s initial product, targeted for consumers, have me so excited? Well, the answer is not what their technology is currently capable of, but what it represents for the future of commercial drones.

When designing R1, Skydio took a unique philosophical approach to what’s currently available on the market. R1 is different than other drones – it flies itself and doesn’t even come with a controller.

Adam Bry is CEO and co-founder of Skydio. He graduated from MIT, was one of the early team members of Google X’s delivery drone initiative, Project Wing, and then started Skydio.

I sat down with Adam to learn about what’s possible right now with the powerful onboard processors and AI that guides Skydio drones, unscathed, through crowded, obstacle-laden environments. We discuss the design inspiration behind R1, a future Skydio platform and SDK, and how commercial drones will continue to evolve.

1 comment on “#070 – Fully Autonomous Drones with Adam Bry, CEO of Skydio

  1. Eric Moody

    As the owner of the first Skydio car license plated vehicle and an R1 I can say with joy that flying this thing is a special learning experience. It’s learning that your whole body is the “controller” , especially when the R1 is leading you. I think it sees the positional plane of my frontal area in minute detail because if I push a shoulder forward it moves in that direction. It’s a thrill to pilot this way, this interaction of positional awareness and a flying object.

    Reply

Leave a Reply

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