The Sky’s the Limit with Chris Anderson (3DR), Adam Bry (Skydio), Laura Major (CyPhy Works) and Arnaud Thiercelin (DJI) | Disrupt SF (Day 2)
2:53AM Sep 7, 2018
We have Chris Anderson from 3DR, former editor of wired, we have Adam Bry from skydio or skydio sorry Laura Major from CyPhy Works and Arnaud Thiercelin from DJI no less to discuss this with Ron Miller from TechCrunch. Big round of applause. Everybody. Come on.
You guys like way over there. I'm gonna. Anyway. Hey. Hello.
So I'm Ron Miller. I cover the enterprise for TechCrunch. And we're gonna talk about drones. Over the last decade we've watched as drones have entered the public's imagination and the industry is just began to develop. First for hobbyists and then as a commercial undertaking, aimed squarely at business.
Perhaps unrealistic uses, sets scenarios have flat led to a hype cycle. And today, questions remain about how the industry will scale and flourish. And we're going to talk to these folks about what it will take for the industry to really take off if you'll pardon the expression. I'm going to let
each of you introduce yourselves, and then we'll get back into the questions, okay? Start Chris?
Yeah, I'm Chris Anderson, CEO of 3DR, which does the drone data. And I'm also the founder chairman of drone code project, which is the sort of Android open source Android equivalent for drones.
My name is Adam Bry. I'm the CEO of Skydio, and we make fully autonomous drones. And we started the company because we think autonomy the ability to trust the drone to fly it so often a useful way is likely to be one of the foundational things that makes all the exciting concepts and drones possible.
I'm Laura major, the CTO of CyPhy Works. We are the leading tethered drone company, which means we have a small cable that connects to our our drone and allows it to fly for hours, even days at a time.
And I'm Arnaud Thiercelin, the head of us r&d, DJI and also head of global developer technologies for DJI and all enterprise projects that are specific to the US.
All right, thank you. So let's start off by looking at the state of the industry today. We saw a lot of hype the initial hype cycle seems to have passed. So where are we in terms of the drone industry?
I probably have the longest tenure here because I started 10 years ago when I was the editor of wired. I started a community site called DIY drones, which was a novel concept the notion that regular people could have drones and kind of the notion that smartphone components could could transform the industry.
Yep, I think that, you know, I don't think the hype has necessarily died down DJI has been tremendously successful, one of the most successful companies in China as a result, I think what you're, what you're seeing is that a classic explosion, kind of Cambrian explosion of different companies and startups all trying to do the space a little bit like the beginnings of the PC industry, and then it you know it more quickly than anybody expected, concentrated around a global giant in the form of DJI and then everybody had to kind of go from the drones to the data side, which takes longer to develop.
Yeah, and I think when drones came onto the scene four or five years ago, it pretty quickly kindled a lot of excitement about amazing concepts, both on the consumer side and in the enterprise. And I think you're right that a big part of the story over the last four or five years has been this gap between the concepts and what you can do with existing products, which is primarily like if you have a skilled pilot that's flying it, you can do a lot.
But the barrier to entry is pretty high, both for consumers and for businesses like businesses, you really want something scalable, where you can just push a button and get the data. And so we're pretty optimistic that, you know, the technology has gotten to the point where over the next five years, we're going to see that gap closed quite a bit.
And, you know, one of the one of the conferences we had this morning, we announced the Skydio autonomy platform, which is an SDK on top of our one, which basically makes it possible for developers to leverage like the full power of the autonomy system that we've developed. And with just a few lines of Python, you can create custom skills you can create your own applications and we think something like that is going to be really one of the keys to making all of the different drone concepts come to life.
guys want to add anything?
Yeah, sure. So about 12 years ago, the drone industry was all about, you know, parts to put together in a DIY set and I mean Chris sure can relate about this and nobody five six years ago we saw the first you know, revolution in the drone industry where these parts
ended up coming up together to create final products, you can just buy off the shelf and stop, you know, working with a directly. And I think right now we are the verge of a another revolution and mean on our side where we've been working on this for for quite a few years at this point and which is very much going to focus on on enterprise me. Of course, we're not leaving the consumer users behind. But you know, in the same point that Adam was saying, you know, we've been building developer technologies for years now to offer all the tools for people to take these drones that we build and start creating all sorts of applications data pipeline in the way that Chris was talking about in a second.
And this is very much because today when you go in and talk to an enterprise customers no longer asking you What's a drone? What's a flight controller? How does he work out as it fly the tell you, I have this problem, you know, this employee of mine as remembering your own personal drones and we saw how you can actually fix it. Tell me how I can deploy 5000 of these tomorrow. This is really one where we are right now. And and all of this is happening because of all these building blocks that happened in the past in the one we have today are definitely sitting up for by future.
So I know that why hood reported that last year the US Army stopped using DJI drones, I know that you fix that problem since then.
And but I want to bring it up as security because if you are launching a bunch of drones, you want to have absolutely be sure that secure their security you don't want the possibility of anybody being able to remotely take them over and do something nefarious with them. So how do you ensure your customers and the public at large that's not going to happen that these things are secure and not not able to be taken over in that way?
So we have I think that's really one advantage of our platform for customers who are worried about security we over our tether we do all the command and control of the drone, and also receive all the data from the drone. And so that offers, you know, differentiator in terms of securing that, you know, there can't be any interference from the outside.
So I mean, I have to pick this up. So first I'm going to be very clear, this memo from the army was completely unfounded on any facts whatsoever, but it did open up our eyes on something that we need to do more which is communicating and very much showing how transparent we are willing to do. And, and after that, I mean, you say fixed it. But the one thing that we fix this.
I mean, of course, we had, we have this ongoing bug fixing that we do every time securities been detected, which didn't start with the memo is not stopping with a memo. But what we did is we hired a private company to investigate all our source code and put an actual report on clearly establishing and explaining what's happening and what's possible and not on the user side. And the bottom line was that we give 100% of the control to user to have their data now on a pure security aspect of things are emotional connection was always encrypted always has been.
But you know, security is never ending game. It's a cat and mouse game. So you always kind of have to chase and fixing. And, and what's important is the response time. What came out of that report, as well as DJI is an average of 48 hours to do bug fixing on anything that is security related. And this is really powerful.
I think one of the like, there's a few different concerns and play with drones, I think that fall in a few different buckets. I mean, I think one of the big ones for adoption is being able to trust the drone to do the right thing. And I think that people's mental model of what drones are today is pretty heavily shaped by this notion of a hobbyist who may or may not know what they're doing, flying it maybe not falling regulations.
And I think one of the really interesting things as these devices become more software defined and software driven is that the rules can actually become part of that software. And so our belief is that as autonomy becomes more prevalent in the market, the model will shift and people will start to just trust these devices more because they'll behave in a more reliable way and you can actually start to enforce the regulation whatever we decide the rules are you can just guarantee that the drone is going to follow them.
it's worth noting our nose too polite to say so but this was and he's absolutely right that the concerns were I think the fifth actually unfounded denial. Think, actually did you I did anything wrong. This is political DJI is an extremely successful high tech Chinese company. And, and we're in a climate in the midst of, you know, trade wars and tear of some crazy tweets coming out of the White House were just being a successful Chinese company is almost enough to spur paranoia.
And you know, I am I wish it weren't the case. But that's the political climate we live in.
But even given that and I will give you that and you have to admit that
if the public has any concern at all that drone could be somehow remotely controlled because it is communicating with the ground, you know, over over the internet, then how do you you know, you have to assure people that that's not a problem.
That actually was not that concerned the concerns about the data coming off the drone getting sent elsewhere?
Speaking of data, let's talk about data.
Regardless of your use case, you guys are all about collecting a lot of data, whether it's photographing something, you know, looking down at a mind security, following an athlete and seeing what they're doing, whatever, whatever the scenario is doing, you have to collect the data and I told this story on the phone. And I think it's worth telling again that I think this was 2014 and 2015, I had a conversation with Aaron Levie from box and he said he wanted to get involved with drone manufacturers and like
Aaron, what do you, what do you want to get involved with drones for and he said, because they use a lot of data and we we store data so when you think about all that data that you're collecting, how do you get that data off of the drone? And how do you integrate it into enterprise systems? So it's not this silo of data that just sitting, you know, somewhere in a, in a, on a hard drive?
Yeah, but that's, that's exactly what we do. It's actually, you know, it's not that hard for a couple reasons. One is that the regulations right now require drones to be used within visual line of sight. So they're actually quite close to the ground station. And often they're used, you know, we use them again construction for example, and they often you know, have other urban areas that might have a connectivity.
So today You know, sometimes it's a matter of taking an SD card out and sticking in the laptop sometimes it's done over Wi Fi, sometimes it's done over LTE because the processing the data caught that takes in maybe an hour or so it doesn't have to be instantaneous Now, obviously be better if it if it took less than then then then that long. And so there's this the push towards what's called edge computing, which is being able to do you know what's photograph material. But basically, you know, computer vision and reconstruction 3d models, I'm kind of on site either in the drone itself, which is essentially what Skydio does, to some degree, or on the ground proximate over over Wi Fi. We do that right now with an iPad.
And, you know, the bounty from Apple's GPU and AR investment means that it's essentially become an edge computing device. And we don't have to go to the cloud until later. So all that kind of real time stuff can be actually done with on the mobile earlier.
So absolutely, I mean, I concur with Chris This is something that can definitely be done on the mobile side. And, and our and our goal is always to create the best platform for people to build application on top one of the components is to create the best pipeline of data that we can create between the aircraft and remote controller. And this is bringing the biggest amount of data and recently with a maverick to we just introduced 10 ADP live video feed and this is you know, continuous work on increasing the bandwidth but there is a problem that actually exist near the data said that you can get from a drone is not on the video, it can be all sorts of sensor data.
And you get an A paradigm where it's very possible to have too much data as we even send it, regardless of the kind of, you know, pipeline that you have down. I'm thinking even five g in his best spec. And so you're confronted into a situation where you need to bring the edge compute even closer to the hardware. So you can actually process that data and at least have a part of the answer you're looking for originally,
because that's just the only way of handling that much data and then you can send that answer back into your mobile and then you can distribute the processing that you want to do and putting it exactly where you want it to do.
So, when I look at any kind of newish technology, I mean, no, no, you guys have been around for a long time. But it's it's new, I think, to most people in the scheme of things. standards have a way of advancing technology, right and I know you're working on a little bit and they provide a base of client services and if you have that base of agreed upon services
I think it advances any industry where that happens. And it's interesting that I was talking to the digital ledger technology people earlier. And they said, you have to be careful that, you know,
do those changes standards too soon, or you can actually suppress the, you know, the innovation that's going on there. So
we're standards that and this industry and, you know, is that conversation going on right now.
it's very similar to the mobile so Apple has their own standard, they're vertically integrated one, they have their own app store their own SDK itself, and I think DJI is very similar to that very, kind of Apple like model, then Android is a more open model. And that's what we do with with drone code. The standards right now largely consists with, you know, sort of common communication standards. There's one called math link, then there's compliance with the FAA regulations and that's something we're all working on together and then the lower level is just regular computing stuff, you know how to how you know, Nvidia in, you know, GPUs talk to
You know, flight controllers and that kind of stuff. So I think you're going to see it play out very much like mobile and word about, you know, we're about like the equivalent to maybe 2010, where we're Android and iOS, where then that's where we are now.
So to extend that point, there's a couple more centers are actually going on that I want to be pointing out. So the first and most important is remote ID, which is a very big topic because, you know, there's a lot of different things that happens on the drone industry, a lot of different scenarios that people are like being concerned about how they can be used. And the first answer that everybody's looking for in terms of government agencies, in terms of law enforcement is identifying what's flying and when it's flying. And so this is a remote ideas is an ongoing conversation of defining that standard across the industry. And we're very much trying to make sure that this is for the best of the drone industry, but we are in a way so big that it were it helps and it doesn't happen at the same time. But this is a conversation that needs to happen with every actor and industry at the same time.
A couple more worth noting. One is around parachute systems on top of drones, the FAA is making one of the requirements to do big loss that you have the capacity when you fly above people to have a safety mechanism, in case you have a critical failure of your platform. And this is a parachute system. And this comes in many colors, what size of parishes what kind of deployment speed, etc. So this is another kind of standardized happening. And the last one I want to point out is around defining what autonomy of a drone is. And this is really, really new, so their groups are getting together and trying to define what exactly that looks like, because that's definitely the next step for drones. Because, like I said, in terms of mass adoption, you know, when you start having a situation where you want to deploy 5000 drones at once, we can't exactly rent a stadium to train them 5000 people to be want to seven pilots, so we are going to go towards this, but just like any standard conversation, they're lengthy, they require everybody to agree to get together and have you know, the definitions.
And the FAA and other regulators are always having the perfect approach. You know, they bring everybody back down in terms of the excitement you have in tech and, and bring you into the reality of what it is because these things are flying next people next to everything. So you do need to have an extra level of rigor when you design anything.
So you guys are tethered. So I was wondering how that how that would impact you.
Yeah, another angle that we're looking at is the beyond line of sight control of drones. So what that means is having the operators be separated, whether that's, you know, miles or away from the actual drone operations because drones allow you to, you know, perform a task that you don't want a person to perform, go into an area hazardous area, watch in an area where you where people aren't going to be there for security purposes. So being able to separate the drone operator physically from the drone itself is really important. And so there's a lot of work to be done in the regulatory environment to enable that as well.
So that you can have a single operator and one command center controlling hundreds or even thousands of drones across the the country or across the globe.
Just to add on that point, all the economic efficiencies of automation come when you kind of increase the ratio of people to machines, because right now, because of FA reacts, it's one pilot one, one round. So we basically have them saved anything in terms of human cost. Once we can go to one too many, then we start to see the economic efficiencies.
So our view is actually a little bit different here. And I think we're really still in the first inning of drones as a category I would say, you know, in terms of phones were maybe closer to like the brick phone that is owned by maybe like a few people around the world. And our view is that the the usage models that are really going to scale that are going to reach 10s of millions of consumers and benefit 10s of millions of businesses around the world are going to be pretty different from what we're seeing today. And so,
you know, I think that there are still a lot of sort of fundamental innovation left to get to the kinds of products
That are going to realize the visions that are out there. So I think, you know, standards are probably part of that. But I think the technology itself needs to progress quite a bit to get to the point where it's really useful for 10s hundreds and maybe someday a billion a billion people. I think that's possible. But we're still very, very early days.
Yeah. You You mentioned regulation. And I wanted to ask about that and the role that regulation is going to play here because I'm curious, do you guys think that regulation is propelling the industry or holding it back or a little both? Or where are we with that?
I honestly think that so far, regulation has been helping the industry for the longest time, if you wanted to have any commercial applications of drones, you need to have a 333 exemption and the pot one to seven has been defining the ground to actually conduct operations. So I don't think that the regulatory bodies actually here to block anything they're coming up with reasonable requests for the severity of the
of the risks that you may be taking with these operations. But of course, you know, when you a little bit frustrated when some of the conversation on being a little bit lengthy, but when you hear their actual case and concerns that they have you quickly realize how they actually necessary to be taken care of.
Yeah, I agree. I think that regular regulation has been a little bit of a scapegoat for the industry. But the trend line is very positive. So before starting Skydio, I was working on project wing at Google X, starting in 2012. And the fear there was that the FAA was just going to totally shut the whole thing down, there were going to be no commercial drones. And step by step, there have been reasonable paths that I think are reflective of the state of products. And I think as the products get better that the regulations get more permissive.
And, and we're pretty optimistic that that's going to continue that.
You know, I think that it's it's worth noting that although I agree totally that the regulators have been remarkably flexible and forward thinking we push them to move faster by using exemptions for consumer use, and getting millions of drones out in the air.
If it weren't for DJI and 3dr in the early days, you know, using these so called recreational flight exemptions, which were not regulated, it wouldn't, there wouldn't be so many people out there thinking about the commercial applications, making the, the commercial requests. So by the time the FAA realized that drones were a thing, we already had so many of them out there that they couldn't shut them, we could put the genie back in the bottle, right?
And I think if you think about it, the government has a role and regulating things that are flying in the air. To some extent, there has to be some guardrails, be some some control because you don't want just chaos, right? And especially if it scales, the way that you guys are talking about scaling. There has to be some sort of sensible regulation on you just don't want to stifling innovation.
So I have a question for you. I was watching my colleague Jenny Craig you talk to the space people have a smart and they were talking about putting these low cost you know satellites, they're very small now, like filling your hands small and they go up into the atmosphere and they collect data and occurred to me that they collect some of the very same data that you guys are collecting with drones, and they're getting cheaper. The you have SpaceX, you have, you know, you have all these space companies delivering them. So as they become cheaper and they become, you know, easier to deploy, is that some sort of existential threat to your industry, or does he do you see them living side by side?
We think about this a lot. We work very closely with a big company called Ezra, which is kind of a geospatial company that started with the dawn of satellites. And we also in our investment, you know, circles, we have satellite companies as well. So, two things to understand. One is that is that just think the, the good the resolution of our imagery is two orders of magnitude better. So, we do sort of centimeter level and they do kind of re meter meter level.
The second is that two thirds of the planet is covered by clouds at any given point in time, and we fly under the clouds. So just from the kind of temporal resolution of being able to go anywhere, anytime get that imagery and the spatial resolution getting centimeter rather than meter, there's always going to be a natural division of labor between those two.
Yeah, I think implicit in that question, is a notion of sort of what drones are capable of today, or have been capable of like, some of the use cases that are working best right now in the commercial space are these ones where the drone sort of acts like a low flying satellite, like it flies above all the obstacles, it uses GPS, it looks down, it just takes downward facing imagery, but there are so many other things you can do if you can actually trust it to fly close to the ground fly through structures fly inside construction sites, you know, the the dream really is it's just this image sensor that's moving around, unconstrained and 3d can be anywhere at any time. And certainly, I think with that vision fully realized there's a completely sort of orthogonal value proposition to just this downward facing satellite.
So one last question is we can almost just a little over a minute left. If there are people who wanted to do a drone startup at this point, and I know you're the closest to, to having just instead of near the most mature as a company, what gaps Do you think still exists that people could explain in this industry?
So I would recommend using the skydio autonomy platform to go
I mean, I'm sort of, there's some really cool stuff that I think can be built, like our system understands the 3d structure of the world around it. And those were people aren't, those were different kinds of objects are and internally we've seen a lot of this come to life over really like the last few months where with a day of dedicated effort, you can build an amazing application for like filming a soccer game or mapping a structure
and there's just a lot of things like that, like you look in any direction and with a little bit of creativity and a little bit of code building on top of the autonomy system. You can do some amazing stuff so
You know, my hope is that we're we're now at the point where all these crazy dreams can actually become reality. And I think there really are some some very exciting applications and businesses to be built there.
I would say the I love their system, and I'm super impressed by it. But I'm required to say, I think they should use the drone code system, which does all that stuff on that quite because there's skydio is really amazing. But a lot of that stuff, it's also open source, and it's free. But more to the point I would recommend they're not start a drone startup and said, they focus on data. There are so many verticals out there that requires expertise and construction or mining or agriculture, whatever we're nerds we know about drones, let you know the drone kind of artist is maybe not entirely solve but it's there's a lot of good solutions out there, how to find it and make these useful how to actually solve real world problems requires expertise in real world problems. And it's not going to happen with the people here on the stage.
All right, we got to end it right there. Thank you so much. Let's give these people a round of applause.