Embedded - 198: Unmanned Flying Thingy

Episode Date: May 4, 2017

Walter Stockwell spoke with us about the legalization of drones, UAVs, UASs, and UFOs. Walter works at DJI which makes the Phantom. They have some jobs open. Aurora by Kim Stanley Robinson Intel/Pepsi drone show at SuperBowl halftime AOPA Facebook page The amateur model aircraft organization discussed was the Academy of Model Aeronautics(AMA).  ASSURE UAS Ground Collision Severity Evaluation Final Report (also: press release) Elecia mentioned the Madgwick Filter. Embedded.fm t-shirts are available for a limited time! There are two distributors: one US based, one Europe based. Choose whichever is closest to you. Elecia’s TV appearance on The Jennylyn Show is on YouTube. Digilent Digital Discovery contest ends May 19.  

Hello and welcome to Embedded. I'm Alicia White here with Christopher White. Our guest is Walter Stockwell and we're going to talk about flying things, probably drones, as Walter comes to us from DJI. First, announcements. T-shirts are on sale until May 18th in the US and in Europe. Yes, thanks to Gert and Teespring, those of you with pounds and euros don't need to pay exorbitant shipping. Sorry, rest of the world. Also, the contest for the Digilent Digital

Discovery units ends May 19th. More details at the bottom of the show, but let's get to our guest. Hello, Walter. Good afternoon. Hey, good to talk to you guys. So we've known you for an impossibly long time, more than 15 years. What? No. Yes, I really, yeah. Yeah. Yeah, I was thinking about the other day. It's a long time. But for listeners who have never met you, can you introduce yourself as though you were on a technical panel? Sure. My name's Walter Stockwell. I'm the Director of Technical Standards for DJI.

I started my life as a physicist and working on dark matter and realized that really had no practical application in the world. And here I was in Silicon Valley. So I jumped ship and started working in startups and found my way into a startup doing aviation type equipment. equipment and ever since then i've been working on uh sort of technical projects dealing with safety and quality and uh making things for people and ended up making things for people to fly with and so that's how i ended up at dji here uh working on that sort of interface between policy and technology that we call standards. And again, focus on making things safe and enabling industry so we can fly things. Okay, we'll have lots more questions about that. But first, we have lightning round where we ask you somewhat random questions and hope that you give us short answers. And if we are behaving ourselves, we won't ask why or give our own answers and opinions.

Are you ready? I think so. Christopher? All right. Favorite movie or book which you encountered for the first time in 2016? Aurora by Kim Stanley Robinson. I've read several of his books, but not that one. It's a good one.

What's your favorite animal? A dog. My dog. Which one? I know you have two, and I bet they're listening. I can't say. Yeah, I can't say in case they're listening. But there is a definite favorite.

There clearly is. Is gravity up or down? Oh, hmm. Um, hmm. Hmm. So, yeah, that's a short answer. I'll just say down. Favorite subatomic particle? You know, I guess I would go with a proton.

Yeah, that's in the nucleus. I guess that counts as subatomic. Preferred voltage um just 12 volts just 12 cartesian coordinates or polar coordinates i usually unfortunately everything i do is cartesian but polar coordinates are such fun should i've gone with spherical coordinates yes i should have yeah you're right it's spherical even more fun have you ever touched a terrapin uh you know that's a good question i don't know no it's not a good question that looks like one i think but was it a terrapin i don't know

we'll go with maybe we'll go with maybe on that. Last one. Drone, UAV, UAS, UFO, or quadcopter? Wow. No. So all of those are synonyms under some circumstances. But usually I just say drone. Drone's kind of tough.

Technically I say UAS. Sorry? Technically you say UAS uas yeah what is the s4 system so unmanned aerial system yep so that includes the ground controller oh they're all part of one system yeah yeah and so and interestingly you you left out one which is our past i've never seen remotely piloted aerial system which is what the international civil aviation authority uses okay but you usually use drones so that's probably what we're going to use today it's easy but it's a very loaded term full of military connotations kind of at their worst. You are right. And so I've been at DJI for a little over a year.

And when I first came to DJI, we were sort of fighting that battle of, you know, is it a drone? Is it something else? And drones have that bad PR. But now even the FAA has given up. Everyone calls them drones. So we're just, we're owning that word. We're taking that back. There's something the FAA, the FAA created something called the Drone Advisory Committee. And this is something that the administrator of the FAA sits on with a lot of other people as advisors. And so if the FAA is calling it drones, you know, who are we to say different? Do any of those terms, UAV or UAS, not RPAS, but do any of them imply autonomous piloting or is that a separate word so that is sort of interesting that's why that's why the that's why our past came into being or

that's why ikea which is the international civil aviation organization that's why they use RPAS, because it implies that it's remotely piloted. So there is still someone who's a pilot. So they distinguish, and so that is not an autonomous vehicle in RPAS. Whereas an unmanned vehicle could be autonomous, could be piloted, could be anything anything so that's why they make that distinction actually are there that many truly autonomous uh flying thingies flying thingies i didn't have that in the list um available for uh industrial or consumer use so legally no legally in the united states or legally all over the world i know you are working all over the world yeah i mean basically all over the world people still um want to have a human in the loop so even so i think we we have some things that are close

there was the uh so intel did these uh the thing for the Super Bowl, right? They had a hundred drones flying around, spelling out things and making the Pepsi logo, an American flag. And obviously there wasn't a person piloting each of those things. So what they do is they call that an automated function. So there's a human pilot that still can hit a kill switch.

And so it's not autonomous because there's a person with some decision making in the loop. But basically, they're pretty much as close to autonomous as you're going to get. They programmed in a flight for them and they hit go and 100

things took off or however many there were and 300 and flew around and made these pictures. Yeah, but they taped them before. They didn't do all that live. They probably had a few tries. Well, so, I mean, we're not invested in Intel and they're not invested in us. But I have met some other guys. And I believe they could have done it live, except for the fact that the FAA is very conservative and didn't want things flying around a stadium full of people.

So they've done these shows before. They've done them in Germany. They've done them in Palm Springs. They've done them in germany they've done them in palm springs they've done them in australia i believe so uh i'm pretty sure what they do is they kind of they plan all the trajectories and they sort of hit go and it goes off and flies everything um and then comes back but there's a person there so it's not autonomous so how far down can you define person in the loop i mean is it just person with the kill switch is it you know somebody it seems like that's defined pretty far down if you're not really controlling it yeah um person who is taking the fall of this all goes wrong a person who gets an email saying oh this isn't working and then he responds to it

five minutes later you know know, disabling them. So here's how the FAA does it. So the FAA says, so first of all, they give you a license for a remote pilot, and they train you. Hey, as a remote pilot, you are responsible for anything and everything that happens to your drone. And they have all these questions. Well, what if someone trips and you're holding the controls? Who's responsible? The remote pilot's responsible.

This is an easy test, huh? Yeah, exactly. They have a bunch of questions that you're like, the answer is always the remote pilot is responsible. So what they do is they say, so first of all, in the rules rules the remote pilot can control one vehicle but they say you can ask for a waiver and and if you show us and you have some method and you explain to us um why you think you can safely control more than one vehicle they'll give you a waiver to control more than one vehicle and so that's what intel is doing there is they've they've basically demonstrated hey

this one person with their finger on the kill switch can control 100 vehicles um using this program that basically automates all these functions uh so yeah so the so the fa has come to the point where they accept an automated function as long as it's sort of executed by a pilot and the pilot has the ability to stop it then they're okay with that and again i think it's because it leaves responsibility with that one person in a way that might be more vague with a fully autonomous vehicle. So Chris has a flying thingy. Jeez.

Several. Yeah. So is he breaking the law? Do you have one of these licenses he spoke of moments ago? I'm not a commercial operator. I do not need a license. I only need to register with the FAA, but I do not need any training. I see. I believe that is correct. Is that now correct? I would agree with that as long as you're flying for fun. Yeah. Yeah. The commercial operators apparently have a lot

more requirements. Mostly it pass a test in which they say, yes, I am responsible for everything that happens wrong in the world. Apparently. Well, the test has a lot of other fun things, like you have to recognize runway markings, which was my favorite part of the test. How often do you need to use a runway? Well, that was my thought. I was like, wow, if I ever see this runway marking somewhere in my drone career,

then yeah, I'm responsible for something things have gone horribly wrong and it's probably not something i want to be responsible for but yeah so they they do have some good stuff on the test um so so you you have to learn about airspace and understand airspace um which is designed to help keep the separation between manned aircraft and the drone so you're not hitting manned aircraft which is designed to help keep the separation between manned aircraft and the drone. So you're not hitting manned aircraft, which is important. There's some stuff about weather, which is interesting because you think, hey, I can

just go outside and see the weather and know whether I should fly or not. And I think that's largely true. But there are some things like, you know, the wind at the ground could be very different from the wind up high. And I think a lot of people... Even not high. 25 feet up, it can change a lot. And around buildings. And, you know, I think a lot of people lose their drones because they don't realize that. Like, it doesn't occur to them until they see their drone.

Kind of like, why is it going that way when I'm telling it to come this way? So there are some good things on the test uh but it but it is it is a kind of scary thing you have to go to an official test center and you have to uh you know prove your id and they they have cameras actually watching you while you take the test to prove that you're the one taking the test. You're not cheating. I know, it's kind of like, wow, that's pretty serious stuff. And you go through a TSA background check. And then eventually you get a little card saying you're a certificated pilot. Remote pilot.

What's the reasoning behind having to go through all of that as a commercial operator versus the free-for-all that recreational flyers seem to be? Because under my reading of the stuff I get from the American Model Association, which represents recreational flyers, even things like the 400-foot flight limit are kind of just a guideline that's not actually a rule you can go as high as you want as long as you're not interfering with with civilian aviation you so that is a very interesting story and um yeah i so how long has the ama been around you know since like the 40s or 50s i think yeah no it's the AMA been around? Do you know? A long time. Since like the 40s or 50s, I think.

Yeah, no. So it's been around a long time. And so they actually have lobbyists in Washington. And they do a good job representing model aviation. And so one of the things that's actually in the law is that the FAA cannot regulate model aviation. Nice. If you're a member of the AMA, you should call your AMA person. I guess you don't have reps.

Go to your club and tell them, hey, good job. Probably if you're a pilot, if you're part of the pilot association for airlines and stuff, you're probably shaking your fist. There's a lot of interesting things behind the scenes there. Because actually, in Washington, these lobbyists, they know each other and they have to work together. They work things out. But this is sort of a weird sticking point in that technically the FAA is not allowed to regulate model aviation. So then the question is, if I have a Phantom 3, what's model aviation? I mean, you could fly that and you're just flying it recreationally, taking pictures, zooming around. I can fly it and maybe i'm taking pictures of a house across the street

to sell to a real estate agent or maybe i'm flying it because i'm developing new algorithms for my company um but it's the same aircraft it's the same drone um and even in the same flight profile it's not like you're necessarily doing anything different, right? Right, exactly. I'm flying in the same, you know, we're both flying around houses and people and whatever, right? So it's sort of the same risk to people in planes. So the FAA had to sort of make, so what they do is they say, well, if you're a hobbyist, you're doing model aviation, we can't really regulate you. Except to say that if you do something so dangerous or stupid that it actually causes a risk to the airspace, that's always a problem. But other than that, we kind of leave it up to the AMA. They don't actually say AMA, they say community-based organization. We leave it up to the AMA and they don't actually say AMA, they say community based organization.

We leave it up to the modeling clubs to self-regulate. And they have a long history of pretty, you know, pretty good safety record to back that up as being a viable way of regulating hobbyists. But you hear all of these horror stories about model planes or drones at airports hitting planes or. Never heard of planes hitting planes. Well, I've heard rumors of such. Really? Usually the plane flyers aren't doing that, but. So, so, okay, so airports, so that's an interesting question, right? Because actually, I don't think there's actually been a case of a drone hitting a plane.

There's been things where they said like, hey, we think a drone hit a plane. And then when you do an investigation, it actually wasn't a drone. There was the thing in London where now they think it was a plastic bag that hit the British Airways plane. Vicious, unmanned, unpiloted plastic bags. Well, that's why they're banning them. And that's why, and I think that's why Chris said, you know, UFO earlier, because it used to be, well, I don't know if you've seen the cute like chart, like, you know, drone recognition chart for pilots. And it has everything from a drone, you know, a quadcopter to a little fixed wing to a plastic bag to a, to Batman. And they're all labeled drone. And, you know, nothing against pilots, but I i mean these things are pretty small they're pretty

hard to see you know i've i've been a passenger uh in an airplane you know in a ga aircraft back at crossbow right where we had to fly avionics and i would sit in the passenger seat trying to run the payload and every now and then they tell you on the radio hey watch out there's um traffic at your three o'clock a thousand feet below you and we'd be scanning around looking like where is this plane and eventually you'd see this thing and it and you'd see it like three seconds before it would if it was going to hit you you know it would hit you um and so i just i have a hard time believing that all these drone sightings are really drones. Yeah, I think there was one in LA where they were, the airspeed of the jet was 250 knots or something.

Yeah. And they claim they saw a drone, you know, closing at equivalent speed or, you know, at a high speed. And it's like, well, okay. Sort of like blink and it's gone. Your eyes must be really good. Yeah. beat and it's like well okay like blink and it's gone your eyes must be really good yeah and if and and you must know chris when you're flying these around well if you don't have your goggles on and you're actually looking at it oh yeah it's pointless it goes 100 yards away you can barely

see a phantom um and something bigger may not even be better because it's there it's not like it's a big flat you know that's mostly carbon fiber rods sticking out and stuff. Right, right. So, that said, I mean, obviously, we don't want people flying at airports in general. I question why people do it anyway. It just seems like the dumbest thing in the... I totally understand why people fly the drones over our sort of shipwrecked boat that's near our beach. Because you can get really pretty beach shots and sunsets and piers.

And then the boat has lots of wreckage and birds. And I can totally see why somebody wants to go fly over there. It's very pretty. Yeah. But an airport, unless you're trying to fake it that you are piloting this thing, that seems like a boring place to fly. So I would probably agree. So I would say there could be legitimate people flying. So one, we do have people who are authorized to fly to airports who are want to fly around to do building inspection or aircraft inspection or even ducks so there i think there is someone doing something

like that like sort of they're running perimeter security and they're trying to run um uh birds away using drones so see there are legitimate uses for drones around airports there's also for example my brother-in-law is so this is in Scotland it's not in the U.S. but actually my two brothers-in-law run a small company doing aerial photography and one of them is actually also a commercial pilot so they needed they they got a a commission to fly right near edinburgh airport it was actually like right a mile sort of downwind of the the runway or whatever straight in the runway and they someone wanted them to go up a few hundred feet and just take a picture of the car lot they were selling cars and they got the

okay to do it so they you know they called the tower and they discuss a method and this and that and so they flew their drone you know right under the approach path of the airport but it was probably at that point the aircraft should be at 2 000 feet um or even a thousand feet so it's well separated so there's a case where like yeah you might,000 feet. So it's well separated. So there's a case where like, yeah, you might actually see a drone, but it's not a problem. Eventually, we're going to have to get to the point where, you know, like when I'm flying into LAX and I'm looking out the window as a passenger, I can see tons of other planes, but they're all lined up getting to LAX.

I'm not worried about it and i and now because i'm working for dji i'm always looking down on the ground to see if i can see a drone because i just think it would be really interesting to see if i could see one um i haven't yet but if it's a thousand feet below you that should just i mean if a plane can fly below me at a thousand feet why shouldn't an inspire or a phantom or you know something fly a thousand feet below me that's going to miss me and so some somehow we have to get to that point does is that the point at which we have trained falcons catching drones eagles eagles right that was the that was the headline of dinosaurs catch flying robots.

So I've seen supposedly, was it Denmark or something, they've been doing that training. France too, yeah. France? Yeah. So I can imagine 10 years from now there'll be a movie where the villain is, you know, Ocean's 27 or something. And they've got this squad of eagles stealing this Amazon delivery package full of diamonds or something. I feel like we should write the plot now. So, yeah.

So, here we go so people are already developed they're using um drones to deliver medical supplies and pretty soon they'll be doing i think someone was actually doing lab samples and so you can imagine like someone having you know the suspected bioterrorism and they're delivering a sample by drone to the lab and the other terrorist like swoops in with his eagle and grabs it and then they can use this sample to infect other people i think there's a tom clancy novel there or something yes so there are people using drones to deliver medical supplies already and i know we've talked on the show before about don't fly your drone near the firefighting efforts because

that's bad for the firefighting planes and fires are bad usually. And I know people have gotten in trouble for that. But I also heard that they are starting to use drones in a fairly autonomous checking patterns to find fires because if you have a big forest you don't always know where the fire is so you're exactly right um and you can get a drone with a thermal camera on it and you can see hot spots um before there's a lot of smoke you know especially if it's kind of windy the smoke dissipates where you can really tell where it's coming from, but on a thermal camera, you'll see it. And I think the thing with people not flying near fires is that, uh, it's not that the

drone itself is bad. It's that someone flying something that's not coordinated with the, with the response team. That's what's bad. Cause then you don't know what that drone will do. So there's actually a lot of fire departments now that are buying drones and are working out ways, how do I coordinate the person flying this drone with the person who's in the helicopter?

Or for CAL FIREs, how do I coordinate the people flying the drones with the people dumping all the water and fire retardant on the forest? Because they realize not only does flying a drone in this situation save money because it's a lot cheaper to operate a drone and see, you know, say you're going into an industrial site. Is this fire near the tank of, you know, whatever toxic chemical they've told me is on the site? In the old days, you either have to get a helicopter up and see it or you just have to send people in to check. And now in two minutes, you can pop a Mavic up even

and just run over and see where the fire is compared to the tanks. So there's a lot of interest in integrating this technology with the existing procedures and processes people have worked out. That's normal situation awareness. I mean, that's important for many, many careers, especially the dangerous ones. Yeah, it really extends what you can do just cheaply and safely. And what we're thinking of now, what you can do with it is not, I mean, people are going to come up with great ideas and be doing really cool stuff with them you know a year from now and two years from now once they start playing with the equipment and really using it in these situations okay so lots of reasons why we want drones and why they are coming they're going to change quite a few things but why are they so hard why can't they be autonomous and geofenced and follow me and long-lasting and safe and deliver hot and tasty tacos?

Choose three, any three. Exactly, right? So there's a lot of pieces of technology that have come together. I mean, it's sort of like an iPhone in a way, you know, there's, there's a lot of technology that we take for granted today that had to come together to make an iPhone. And, um, when you look at drones, it's sort of a similar thing, like, uh, batteries, battery technology has to find a high density energy, know uh form factor right i mean it's hard to beat gasoline which is what we've had before um the sensors that go in to make it stable um like inertial measurement units exactly so those uh so those i mean because of games and

smartphones like those are super cheap and actually super good these days. Do you remember when we made like the first one of those and it was big and it was so amazing to have a MEMS IMU and people were just gobsmacked and why aren't you using fiber optic gyros and this is never going to work. And we thought it was pretty good when it was $100 in access. Oh, yeah. And now it's like, yeah, okay. You know, $3 for all six, that's too much. I need $1.85. And no gyro-bio-strift, because that's just for suckers.

Don't give me that. It's all surface mount part, the of it that was that developed really fast i mean because you guys were doing that i feel like 10 years ago now that was 15 years ago 15 all right yeah still time flies but but yeah i know there's so the i mean we had to hand calibrate each of those oh and that was like a dance you had to put it on each side with these really amazingly flat surfaces. And I could never do it. And I remember Tony and he would just like do it really smoothly and swipe it over and then put it this way and then Z and then X and then X of things. And now, I mean, these things, they're bulletproof, practically. And so that's a big piece of technology that's made. I mean, that's how DJI got its start, actually, was helicopters were so hard to fly. And our founder was like, I'm going to make a flight controller that can make it easy to fly. But so then when you think about what makes it hard, I think, is that – so think about autonomous cars.

So those have been going around for a while. And at least in a car, you've got a huge frame, a gasoline-powered car. You've got all the energy you could ever want. You can support cameras, LIDARs, any kind of sensor you want. Now, to make that fly, you could make a big thing fly, but that's really scary because you can't just pull over. What goes up must come down. Right. If something's not going right, you can't just stop.

You're in a car. You can just sort of pull over and check on things and now you have something flying you can't just pull over um and i think that's the hard thing to me that seems like the hard thing is you just can't pull over when you're flying so um and now you have a drone let's say you want to do deliver hot tacos over to the beach um there's seagulls flying around there and there's other people flying drones and um maybe i want to avoid the airport on the way over there there's just a lot of things to think about um and it's and it's not easy to figure out. So airplanes, you know, fighter jets use radar. Commercial airplanes use a combination of radar on the ground and radioing each other their position.

But they're basically hoping there's no birds up at 30,000 feet. Which is usually reasonable. Which is usually a good, yeah, right. right i mean but now i'm flying at 200 feet you can't assume that so you have to have something that can find the birds can find the emergency helicopter flying through and they're going low and quick because they're trying to save someone's life um you have to avoid buildings and trees and power lines and power lines are hard to see you know whether you have a radar or a visible camera yeah airplanes don't usually have to deal with power lines yeah not at 30 000 feet right so there's just i mean

it is a hard problem and then it's just hard because it's hard there's a lot of there's a lot of things to think about and you can't stop and pull over and think about it and you and if you make a mistake it's gonna fall down and you'll lose your tacos the tacos will be late and you'll you'll get a bad rating for your taco delivery app or whatever. Bad Yelp review for your drone service. Exactly. Why do people talk about delivering tacos? It makes no sense to me. I think it's just because it's the most ridiculous application.

Oh. Isn't it? All right. Yeah. Well, it makes no sense to me. That's where I was headed. But even Amazon deliveries don't make sense to me.

Well, it's because we're old. Well, no, it's because we're old. Well, no. It might be. He was in middle school. And I was like, yeah, I was over there, you know, watching them deliver burritos or, you know, practicing delivering burritos. And he's kind of like, huh?

And I was like, would you like, like, what do you think of that? Would you want someone to deliver you a burrito by drone? And he's kind of like, huh? And then I was like, would you like, like, what do you think of that? Would you want someone to deliver you a burrito by drone? And he's like, sure. And I'm like, what if it costs more money? And he's like, well, how much? And I was like, I don't know, $5 more to get a burrito. And he's like, yeah, sure. That's fine.

He's still spending your money though. Yeah. So maybe that's, maybe that has to do with me teaching him value. But, you know but it would cost more because putting something in the air is uh less power efficient and therefore is going to cost more than putting it on wheels i just don't think that air air delivery can ever be cheaper well and so you're right so it's not can ever be cheaper. Well, so you're right.

So it's not going to be cheaper. So they can't get you on cheap. It's going to have to be on quickness and convenience or I need it. I don't know. I'm not one of those people who wants one hour delivery from Amazon. But I'm not the market of most of the people marketing, I guess. Well, I think the plan is, you know, the plan is to have lots of distribution centers for common stuff, right? So it's not going to fly super far.

It's not going to fly from Sacramento to San Jose or Nevada to wherever, you know, they're not going for the big shipping. So yeah, I don't know. I mean, for stuff like, oh, God, I really need, I forgot to get, you know, whatever. If it came in an hour, that'd be pretty cool. I think there's a big market of people who would do that.

I mean, I know you would get things delivered like three times, but then after that, once the novelty's worn off, it wouldn't be that exciting. I'd be doing it every day. Amazon Prime. There's your market.

Now, flight. I'd just be getting a pencil three times a day. A little drone would knock on your door. Forgot the eraser. Okay, we've gotten so far off the topic I actually wanted to talk about. I'm sorry. I wanted to talk about the legislation because that's the thing that's changing. Drones are coming and they are finally to the place where

we have all the pieces or the pieces are coming. I don't know if the batteries and smarts are quite ready, but it's definitely headed in that direction. But how do we make sure that these are safe? How do we make sure that they don't hurt people. There's a lot of background for the FAA and even civil aviation making sure that you at least understand how not to crash your Cessna into somebody's house. And all of the hardware and all of the documentation that goes along with that. But crashing your drone into a house may not cause as much damage

but it's still not going to be trivial what are you doing with respect to the legislation so that's a great question and it's um you know depending on how you look at it so the fa treats these as aircraft and the way it regulates other aircraft is it is it there's a safety goal and the safety goal is in fatalities per hour of flying and uh if you want to have sort of a morbid if you have a morbid curiosity you can you can go on Facebook and join the group for the AOPA. So Aircraft Owners and Pods Association. And there's basically a small aircraft that goes down every other day. And people are dying.

And we're trying to set a very different kind of safety standard for drones, which I don't think is bad. But it is a very different standard. We're now thinking, OK, because there's no one on the drone, you obviously don't care. You know, unlike for GA, if the plane crashes, there's always someone involved and they're almost always going to die. But then with an unmanned system, to first order, sort of who cares if it crashes? It's just money. So then you have to sort of look at these second order effects. And those happen to be, like you said, if it crashes on people, it could hurt them. If it hits another airplane with a person in it, that would be bad. That could bring down that plane and that person in it. So governments around the world have taken a pretty conservative approach to this, which is, one, they just started out saying you can't fly these things.

We're calling them aircraft, but there's no legal method to fly them under. Yeah, we did that with the FAA. But that situation has gotten better six or seven months ago yeah and then they finally they did this part 107 and they they now have this way that we talked about earlier about getting a license and then you can fly it commercially but in a lot of those rules though are are basically basically everyone's idea is that when you fly you should keep it in your visual line of sight. And that has two purposes. One is that if you know where it's flying and you can see it, hopefully you're not hitting obstacles.

And hopefully you're not over a crowd or where if something happened, it would fall on people. But also, you're supposed to be looking around the airspace around you so that you could see if a helicopter is going to come flying by. You know, hopefully where we're flying, no one's supposed to really be flying up at 5,000 feet. So there shouldn't be other aircraft that you can't see. But, you know, you're supposed to be able to be responsible and keep this from hitting other people or other planes. So that's the basic legislation that most countries around the world are implementing. It's this visual line of sight flying. But so many people are doing the first person view where you wear cameras. Right.

So that's the flip side of this really is that if you look at like the faa registrations um i think there's about 30 000 registered or certificated pilots remote pilots right so there's 30 000 commercial flyers but there's more than 700 000 registrations so there's a huge number of people flying recreationally um and dji that's one of our largest customer bases so we know this there's a huge number of people flying recreationally and they've been doing that for a few years now so there was a big change if you were flying 10 years ago you weren't flying a little quadcopter you were probably flying a fixed wing aircraft and you're at a park you know flying in circles um maybe doing acrobatics uh some people have these helicopters yeah you're not doing first person view and you're not flying

these little quads and you're not sort of just going out in the world and filming things you're you know you're a part of the ama they had flying events you were at club sites um but now for the last few years we have had all these quadcopters and people are flying them all over the place and they're they may or may not be part of ama um but we have you know millions of flight hours of people not hitting planes and not hitting people um there's very few incidents of people getting hit by drones. So that in itself is sort of evidence of the operating threat from these, I guess you'd say, or, you know, how dangerous these are. And that doesn't mean to say we shouldn't have rules and we shouldn't have laws about this but but i think people are you know if you take that seriously that there really are

millions of people flying millions of hours flying with no serious accidents then you have to say well we're we're as safe as manned aviation in a lot of ways. And that's part of the reason why not only in the U.S., but also in other countries, they've recognized their model flying communities as being safe enough or not requiring additional legislation to make a safe flying environment. Is this sort of akin? If that makes sense. I mean, it sounds like it's akin to when cars first came out and there were laws saying that you had to have a flagman walk in front of the car. Although, to be fair, we are talking to a DJI guy about legislation for drones. I suspect that the company doesn't want a lot of legislation. So, no, that's an interesting question.

So here's what I would say about that, is we want a safe industry. If that requires legislation to make it safe, we would support that. And I think so we do actually have a global policy drones are operated, both commercially and recreationally. We've been world leaders in setting up geofencing systems, for example. And we so our global policy guy sits on that drone advisory committee that I was talking about working with the FAA directly. So it's not that we're against legislation. We we want smart legislation, I guess I would say. And smart doesn't mean something that gives us an advantage or just means people can fly wherever they want. I think smart means it's appropriate to the level of safety needed and appropriate to the to the risk presented um and so actually one of the things we've been doing is we've been supporting research

into this very question of how dangerous are these drones and the faa more than a year ago set up something called ashore which is what they call a center for excellence for drones for uas and we've been supplying them with materials and advice for you know hey let's take drones and throw them at crash test dummies and see what actually happens. How dangerous are these things? I'm surprised that had never been done before. I mean, the hysteria came well before even assessing. That doesn't surprise me now. Yeah, keep talking.

It all starts to make sense once you hear it aloud. Yeah, no, see, here's the thing, though, is that you think about it, manned aviation, if there's a crash, like, you don't need to need to like have a crash test dummy in a plane to know that's a bad thing right so from manned aviation's point of view um there wasn't this you know there's not sort of the same kind of history behind this they're just like a crash is bad like any crash is a catastrophic event um and so now we have to kind of recalibrate that. Like, what does it mean to be a catastrophic event with an unmanned system? You know, if I crash in the desert, who cares? You know, if I crash a small drone on top of my house, who cares? And so, and it turns out crash

test dummies are kind of expensive. So it really took sort of a push to say, OK, we've been assuming it's bad if it hits a person. And obviously it's something you don't want to happen. But we need to know, is it really, you know, is it going to just kill them? In which case we really should have pretty severe legislation. Or is it, you know, like getting hit by a soccer ball in the back of the head when you're not expecting it? In which case, like, yeah, you shouldn't do that. But it's not like we need to lock down the entire industry to prevent it. And they've just released their first report on Friday. It was pretty interesting news.

If anyone's interested, you can look up a sure ground strike report. And they did a really interesting experiment. So they did a bunch of experiments of hitting crash test dummies with drones. And they're comparing them to federal motor vehicle standards for airbags and crashes and things like that, where they have a pretty good correlation of what the crash test dummy measures and what injuries result. But they did this one thing that was really smart and really illustrates the difference between a drone and something else, which is that they dropped a drone onto the crash test dummy's head and it bounced off and they measured everything. And then they took a block of wood that was the same weight

and they dropped it on the dummy's head and measured it. And then they took a block of wood that was the same weight and they dropped it on the dummy's head and measured it and then they took a block of steel the same weight and dropped it on the dummy's head and measured it and what they found was drones glide no well so they hit at the same speed so it's the same kinetic energy same mass but the drone flexes around and cracks and bounces springiness yeah and so the crash test dummy basically it said there's zero chance of a head injury and maybe a 10 chance of a neck injury by the federal motor vehicle standards but you drop a block of wood at i think it was like 50 feet per second um they're like yeah that's pretty much 99 chance of a very serious head injury you drop a piece of steel like yeah that you're gonna really really hurt that person what about the rotors i mean i think a lot of people are afraid

of the rotors and eyes which i guess is sort of like being afraid of birds and eyes, but I blame a certain movie maker for that. So you're right. I mean, I would get nervous about rotors and eyes, and I'm not surprised people would. And so that's kind of one of the next, I think actually there was a little bit about that in this report they just did for Ground Strikes, was-

Ballistics gel, I imagine. I mean, you could totally use ballistics gel i think mythbusters did this did they oh that may be why i have it in my head as a picture but also they have these things so it turns out so wow i mean so i've been doing so much research into this where like i'm reading reports of tests on cadavers and they have pictures and things it It's stuff like, this is why I went to physics is because I didn't want to see things like this. You know, and not biology. But they have, so it turns out like ballistic gel, like it matters. Like, does it have a skin? You know, does it have, so there's actually fake skin you can get, skin simulant, you know, you can get as well.

And so for the rotor test. That's going to be really fun at your house isn't it yeah i mean i think they so that i think i saw um tests they did with like skin simulant and on top of you know rib pork ribs and things like this and then putting a motor on a pendulum because because it's all you know if i just take you know a motor in my hand and start the prop spinning and like push it at you yeah i'm gonna hurt you probably right but if the drone kind of bounces against you that's a different thing um so there's just a bunch of these things that make complications and i think at the end of it it was like yeah if you could hit but and it matters the size of the drone so something like um you know a little fpv racer um with propellers that are probably only an inch and a half or two inches. I don't, I don't know.

You guys probably know better than me. Yeah. I think the ones I have are three, three inches wide or three and a half. Diameter. Something like that. And then you look at like the big, like we have something called the M 600,

which can lift, I think 10, 10 kilograms or something. Wow. And, um, yeah, and it's a big, we deliver babies with that.

Yeah. You good. Two twins. And, yeah, it's a big. They deliver babies with that. Yeah, you got twins. Big twins. Delivering babies. Yeah, that would be a good marketing thing, right? Oh, the DJI stork. Those things, each blade is like a foot long. Yeah. And I think in this report actually had something about it and they're scaling and you know basically the longer the propeller the more dangerous is going to be for

a lot of reasons probably has to be a stronger propeller it's going it has more energy you know etc etc um but even a drone with a foot long propeller is probably not going to chop your head off. It's going to really cut you, you know, that'll be bad. Um, and, and that should be prevented probably, but it's, um, it's probably not still, it's still not as dangerous as like a car hitting you or um a plane hitting you obviously or that eight-year-old on the razor scooter that drives by really fast right i mean so there's so it's i mean this is kind of hard to talk about right because it's you're talking about risk and no one wants to imagine the worst thing happening and say that's acceptable, right? But we do that all the time with kids riding bikes to school and cars driving on the freeway.

I mean, you could have geofencing for cars where you have a GPS in the car. They wouldn't drive into post offices, such as the car that drove into our post office. And they could have speed limits. Right. And we don't require that. Yeah, it's these grandfathered risks that we sort of slowly accepted like cars. I mean, if cars were invented today, I can't imagine what the regulatory environment would look like. You probably would have a flag man.

Right? And so, like you asking, like, well, what about, you know, getting cut or hurting your eyes, you know? And now imagine applying that to cars. And by saying that, I'm not trying to say it's okay if these drones, like, cut you to pieces, you know? No, but we have frameworks in place. But it is a different kind of safety we're talking about. I mean, we have insurance. We have different kind of safety we're talking about. I mean, we have insurance. We have civil lawsuits.

We have things in place to deal with the risks we take every day and the risks that we inflict upon each other. And that's why we have to have car insurance in California, because you have to admit you're inflicting a risk on those around you. So an alternative would be, so I guess a parallel to that then would be to say well we should require drone insurance um as opposed to just saying these things shouldn't fly right yeah but we're still at the we're still at the point i think where nobody would be sure right now yeah i get i get a quarter million dollars of insurance for being an amm member which is a good reason to be an AMA member. Yes, indeed.

For any of those listening, you should look into the AMA. It's a good organization. Plus, they have excellent lobbyists. But it is a different kind of risk. It's a new thing. And I don't think it's bad, necessarily. We should have higher standards. We should try to make a safe product and a safe industry as this comes in.

But it is shifting the goals a bit. You know, you know, you can go, you can build an experimental manned aircraft and fly it. And no one's going to worry about whether you cut yourself on it, right? You're either going to kill yourself or you're not going to kill yourself. Okay, I'm going to shift topics and it's going to appear weird and odd, but I promise this comes around. What is a FAA designated engineering representative, a DER? So here's what a DER does. The FAA has design standards for aircraft, for example.

So let's say I make something that's either going to go on an aircraft or I make an aircraft itself. I'm supposed to be following certain design requirements set either through the federal regulations or through TSOs. And we talked about DO-178 and its different levels here before. So that's one form of the standards that they have. So design standards could be software or hardware. But here's the thing is the FAA doesn't really have the people to check up to see if you're actually following those. So they outsource it like any good American company. You get contractors. And here's the,

here's the cool thing for the FAA is that they don't pay for the contractors. You, the person building the thing, pays the contractor. So it's this weird half and half thing where the FAA finds someone they trust to be honest and proficient in the kind of thing they're inspecting. And so they designate authority to that person. So that's where the designee comes from, the DER. They designate a piece of the FAA's authority onto that person to make this judgment of whether or not you're following the engineering standards. And then as the applicant, the person making the design, you hire this DER to just sort of look over your shoulder and make sure you're doing everything right along the way. And at the end of it all, they put a big stamp of

approval on it. And the FAA then puts their stamp of approval on it. And so a DER needs a fair amount of training and both in the technical pieces and in the certification process and some sort of ethical training? Yeah, I mean, it really is an interesting space. So you usually get people that have been in the industry or perhaps in the FAA as oversight. And so they've gotten this training kind of organically. And then they decide for whatever reason they want to go out on their own and they sort of become a consultant and do this DER stuff. Which is not cheap, by the way. Yeah, well, but it's so valuable. Well, you're not going to get the FAA's attention. If you get a DER, yes, you're going to pay for it. Otherwise, you're not gonna get the faa's attention if you get a ddr yes you're gonna

pay for it otherwise you're gonna wait five years and the um but you're right also there's this moral dimension to it or an ethical dimension where you as the designee have to have a relationship with you actually have sort of a mentor in the FAA or I forget the official word, but there's, I mean, there's, there's a caseworker sort of assigned to you and you have to go to regular training and they kind of look over your shoulder.

And usually when you start, you actually have another DER as well that you're kind of partnered with to kind of parallel track and sort of work on the same project together and, and make sure you're kind of thinking about things the same way. And the, so here's the other interesting thing though, is that as the DR, as the designee,

you have the authority of the FAA, but you don't have the protection of the FAA. Do you get to wear a sidearm? No, not, not. So, and, and, but you, you it's funny but so there's so the fa as an fa employer federal employee you're not personally liable necessarily for anything you

do or decision you make you've got government lawyers behind you but as a designee you're not a federal employee so you don't have an army of lawyers so you have all the authority but then you're also hanging out in the wind exactly so so that is kind of the stick part of that um so yeah it's you know maybe that's why they're expensive as well um you know you're really on the line for everything you put your signature to. It's impetus not to screw up. Right. And a big part of the training is them telling you, like, you don't have to sign something if you're not comfortable with it.

You know, say no. And you know this from personal experience. When did you get your DER designation? So I wasn't actually a DER. I was a DMIR. Oh, my God. Slightly different. Which is a designated manufacturing inspection representative. Ah, to make sure that the units were coming off the line as they should be instead of all sorts of calibrations randomly done by one person in his cubicle.

Exactly. So I was signing airworthiness certificates so yeah so i was so i was taking responsibility so so how that happened was um back at crossbow we were making an fa certified device and at the end of it you you have to get an airworthiness certificate which is the fa saying um we agree that this particular equipment was made to the standard that we approved and is in good condition. So you can pay someone to come in from outside the company, and that's very expensive to sign every single thing that comes off your line.

Or the FAA has a special thing where they let you, if you're an employee of the company, you can become a DMIR. And then you again have this special hat you wear where, you know, when I put on my DMIR hat, I'm working for the FAA, I'm not working for Crossbow. And so I have to be able to say no if I don't agree with this thing you're putting in front of me. If I don't agree it was done right, I can't sign it. So, yeah, that was kind of a fun thing. I did that for about seven years. What a weird idea. It's like you're a government mole.

Everybody knows it. You know, I think about, you know, like it would make a good novel or a good science fiction, you know, just the weird quirkiness of the legal system and the crevices, just the way things work. Yeah. So how has it been different for you working, making things that the FAA regulates and now having a role trying to tell the fa how to regulate things so it's actually i think it's really interesting um and i think my experience before helped me in that i've seen um so one actually coming from a thing where you're making avionics for manned aviation i mean mean, we had customers who died flying. Not because of our products, thankfully. Get that in there. Yeah. But yeah, no, I mean, like, Crossbow took things very, very seriously in terms of,

if you do something wrong, people can die, right? And so now here i come into this company dji which really is coming from a different kind of background that's more consumer electronics i would say um because they were making things that were small and fun and anyone can fly them and and got super successful and now it's an issue really you know if there's only five of these things no one's going to care but if there's five million you know people are going to care um and so i i'm i'm able to be kind of a bridge between the old aviation culture and this new um innovation culture and dji has been very open to that um i've i'm really enjoying working here for that reason, that they take that very seriously. And like I said, they want to have a safe industry.

So working with the FAA, I'm talking to these people that are coming from this very traditional aviation background. And there are some people in the UAS integration office. Well well actually, so I should say they've been very supportive in general in that, um, they do, it seems to me that there's a lot of people there that recognize that unmanned systems are going to change the way aviation works and they want to make sure that happens in a process and not just be overwhelmed by it, which is kind of what started to happen before they had this part 107 people. I mean, there's a million people flying and there's no real regulation for how to do

that. Yeah. Regulating and saying you can't is not regulation. Yeah. That just, that doesn't work. Yeah. And,

and I think these, you know, the FAA recognizes it and not just the FAA, I think the rest of the aviation community. Back a little over a year ago, when I first joined DJI, I was on something called an aviation rulemaking committee for figuring out how to fly drones around people and over people. And so that was a really intense two-week thing of sitting in Washington, D.C. around a table with 30 other people. And there was someone from the AMA, but there was someone from the Airline Pilots Association, someone from General Aviation, someone from Agricultural Aviation, aircraft executives.

So it was trying to be like a pretty 360 view of the people involved in aviation and people that would be affected by drones flying around. And so I'm sitting next to someone from the Airline Pots Association, and he's telling me, I'm really worried if this hits a jet, it's going to kill a pilot and 100, 200 passengers. And, you know, that may be right. A big enough drone hits in the right way that could bring down a jet potentially. You know, that's something we should find out. But it's a really interesting process of watching how all these different stakeholders with different points of view are trying to bring their concerns to the table and kind of trade off you know like what what am i just irrationally scared about what am i realistically

scared about um and i'm doing the same thing i'm trying to say you know like yeah if that hits a plane that's gonna be bad for us it's bad for everyone on the plane we don't want that to happen either um so yeah i think i've just lost my train of thought that's fair i i took you on a bit of a random ramble. So, so don't worry about that. Let me ask you a couple of questions I got from listeners. This one's technical when flying a quad copter and there's a prevailing wind. Do you use accelerometer gyro GPS or a combination in order to keep a

predictable pace? is this where a kalman filter comes in yes exactly so exactly all of those yes yes yes so uh so i mean it's a huge question depending on like what you're flying and why you're flying but in general yes yeah that's exactly um you would want to use all those things to figure out because you and you should be able to estimate basically um you know what's the wind you're going against based on you know how much tilt did i have to apply to get that speed um you know what's my actual ground track versus my predicted ground track based on how I'm tilted and things like that. So yeah, you can make it simple. You can make it complicated.

So a Kalman is a way of doing sensor fusion so that you can use your accelerometers to decrease your gyro bias and your gyros to decrease your accelerometer noise. And they're pretty tricky. I mean, they're kind of hard to calibrate. And the more sensors you have, the more accurate you can be, but only if you really understand how all of this works. There are some different filters out there. Are you familiar with any of them?

Well, so I think the filtering method is different, I think, from understanding. So, I'm a physicist years out of the system design stuff but um there's a lot of you're right there's a lot of different filters but to me those are just ways of optimizing the computation power or optimizing the kind of solution you want but it really comes down to what are you physically measuring and what is that what are you able to derive from that so if i just had accelerometers for example um as you know you know i can't really tell motion a change in my speed from gravity for example so i can't use that solely for orientation if I'm moving. And some people don't understand that.

That's a basic physical concept. And I don't know if you had to talk to many customers at Crossbow, but some people didn't understand why you can just only use accelerometers because gravity is always down. It's not labeled well, though. Exactly, right? It's not a different color. That's kind of how I'm like, yeah, it's not a different color.

Acceleration is acceleration. It doesn't come in different flavors. Sure it does, X, Y, and Z. But you have to start. You have to know which those are, right? Or it could be theta and R, R dot. But to me, so like a common filter versus a point cloud filter or some other kind of filter, it's really,

you have to make a model first of what are you physically measuring and what do you think that does? And if you don't get the model right, the rest of it doesn't really matter fair enough i came across magic filters lately which magic i was it was kind of cool okay magic magic okay oh i think you said magic oh no sorry magic would be much better um magic was just a lot of math that came out to be not that far from... What's its claim to fame then? You got somebody who just...

Probably. Exactly. Less tweaking. It was only one knob to turn. Okay. And pretty light calculation. The computation was not as heavy as a Kalman. Who wants only one knob to turn?

People who actually want to use these things. Burrito deliverers and things. Yeah, exactly. Okay, second question, which I actually have heard the answer to already from Christopher. Could you please change your GPS lock LEDs from green-red to blue-red since 16% of men are red-green colorblind?

This was nearly shouted even though it was IMU or even though it was instant messenger Slack really. It was nearly shouted at us. We'll see if your answer is the same as my answer. Yeah, Chris had an answer, but let's see if Walter's answer the same as my answer yeah chris chris had an answer but let's see if walter's answer is better so now oh now i'm now i feel on the spot because i mean

i i know plenty of people who are colorblind i think that's a great idea i mean there should be some way of um well i don't know it should be but i mean wouldn't that be a great feature to have the ability to control the colors of the leds on your thing mine mine does mine's got a whole row of super bright leds round effect leds i can make it like do pac-man on the back while you fly or whatever uh my answer was that that red and green are the standard aviation wingtip flasher colors so you can tell that so that's why I thought they chose that. So that's certainly why. So, I mean, that's sort of traditional, right? But here we are. We don't have to follow that tradition.

We got rid of the flagman in front of the car. It has to do with what the FAA tells us. But, you know, so they talk about conspicuity. And if you're flying at night, you have to have strobes that you can see for three nautical miles and things like that but they don't actually say you have to have traditional lighting on it so i think that's a really interesting idea and um i'm gonna take that back and ask can't i don't i don't have any i don't really have much leverage on what actually gets built so i can't promise your listener anything but i think that's a good idea.

Well, we have kept you for quite a while and I have more questions. Christopher, do you want to jump in or should we just let him go free? I think I asked the ones I had queued up. All right. I do want to ask you about jobs at DJI. So that's a great point. We've been expanding. We do have offices worldwide. And I'm in an office in Palo Alto, for example. You can check our website,

and we should have a careers section. Take a look. We're always looking for great people. And if you've seen our products, you can see we love making great products. So it's a cool place to work. And I heard the answer to this recently as well. But I know everybody who's looking at a job at DJI has the main question of how much is your employee discount? We don't need to go into that. We don't have to go into that. 20% off coupon just for interviewing?

Maybe I shouldn't say anything, but I will say there is an employee discount. And actually, there's an interesting thing. So the first time I went to China to visit headquarters, I was told that every engineer in China is expected to learn how to fly the things. And once they prove that they're proficient, they're given a drone and told they have to go fly every week. Oh, no, the horror so that hasn't happened in america yet i haven't been given a drone but um yeah we're we want people to fly we it's best you know go out and fly these things that's the best way you're going to understand what you're making yeah it looks like the jobs in palo alto are robotics engineer which if you've been listening to the show, you know what most of those acronyms probably mean on that page.

iOS developer and Android developer. So there were others, but those are all we care about, right? Right, right. Yeah. So do you have any thoughts you'd like to leave us with or questions we should have asked you? No. with or questions we should have asked you uh no um but you know now that i've done this maybe you know a year from now i'll come back on the show it's not so bad um so my only thought really is that uh i think there's a lot of great things drones are going to do in that just like other automation,

you know, other industries where automation is coming in, where we're going to make things safer for people, where they're doing jobs that are dangerous or dirty or distant. So hopefully we'll get past this period where people are afraid of drones and people will start to see a value there. I think if you look into what they're doing, we've been tracking reports of drones saving lives, for example, and we can point to 57 lives that were saved by people using drones

and there's been zero people killed by drones. So we're already on a good, positive upslope. And so that's kind of the thought I would leave with people is to think about the good things that this technology is going to do and keep an open mind about that. All right. Our guest has been Walter Stockwell, Director of Technical Standards at DJI. Thank you so much for being with us, Walter. Thanks. It was fun. Thank you also to Walter's wife, Emma, for lots of things, including black flip-flops.

In somewhat related news, my TV appearance is on YouTube. What happens when you turn all of your nerves into weird intensity about embedded systems? Well, let's see. Also, last week we talked about the analog discovery because that was what we opened for ourselves, but then we got to reading about the digital discovery, and it goes a lot faster. Eight channels at 800 mega samples per second as well as 100 mega samples per second output, 16-bit pattern generator. We should have read more about it.

It was very cool. And why are we giving them both away? Why didn't we keep one of those for ourselves? Yeah, I don't know either. But we are giving them both away. You're supposed to send me a number from 1 to 997 inclusive and hit the contact link on embedded.fm. That's the best way to enter. And note that one will go to a student. So tell me if you are a student so that you get the extra, you know, dual entry.

And now finally, thank you to Christopher for producing and thank you for listening. And a thought to leave you with this one from Wilbur Wright. I confess that in 1901, I said to my brother Orville that man would not fly for 50 years. Embedded is an independently produced radio show that focuses on the many aspects of engineering. It is a production of Logical Elegance, an embedded software consulting company in California. If there are advertisements in the show, we did not put them there and do not receive money from them. At this time, our sponsors are Logical Elegance and listeners like you.