Original subtitles

[narrator] On this episode of How Tech Works...

we get exclusive access

to the latest military transport vehicle.

It's as versatile as a passenger van,

but it's built for the skies, not the suburbs.

And, we'll meet a robot that can predict exactly

where a moving object will land.

Sounds like a cricketer's dream.

[show intro]

[futuristic music]

Hello there! I'm Dr. Basil Singer

and you won't want to be anywhere but right here

for the next half hour, as we explore

the latest advances in science and technology,

and awesome robots that do cool things.

I'm talking a real, life-sized robot car

that can drive itself, down the Autobahn!

And, I'm talking a vertical wind tunnel

that lets you skydive indoors!

But first...

We here at How Tech Works like things that are big.

Really, really big.

So we thought we'd take off with a massive plane

from Seville, in Spain.

It's called the Airbus A400M,

and it's one of the world's biggest air transport planes.

Here's our exclusive behind-the-scenes tour

of what makes it tick.

[rock music]

[narrator] You're looking at a test model

of one of the most massive planes ever.

This is the A400M, designed to challenge

the air transport planes that currently rule the sky.

At 45 meters long, it's a whale of a plane.

Bigger than an AC-130 Hercules.

But like the Hercules,

it's designed for both military missions,

and humanitarian aid.

It can even serve as a petrol station in the sky,

refueling other aircraft in-flight

or taking fuel onboard.

One look at the cargo hold says it all.

The cargo hold has a maximum capacity of 37 tonnes.

That means we can transport a wide variety of things,

like skydivers, paratroopers, medivac,

even helicopters, without rotor blades, of course.

It can drop cargo at a low speed and altitude,

and land on short runways without any prior preparation.

[narrator] It's one of the most powerful transports

on the planet, with four turboprop engines

providing a total of 11,000 horsepower.

Okay, so it's a monster. But what's it like to fly?

It's a big truck, but it flies a bit like a sports car.

[narrator] Test pilot Tony Flynn has taken it out

for a few test drives.

We can go to 120 degrees angle of bank,

we can go to 3G,

but of course, with that comes responsibility,

because it doesn't matter how good the sports car is,

you can still drive her off the road.

[narrator] That's why the mechanical controls

are different than in a car, or even an older plane.

[Tony] In most old airplanes,

the joystick at the front is connected mechanically

to the flight control systems on the airplane.

What we've done is we've removed that mechanical link

so the pilot is connected to a computer.

[narrator] The computer system is called fly-by-wire.

It's programmed to guess what the pilot wants,

but to fly within safe limits.

[Tony] There is a G limit,

so he cannot overstretch the airplane,

he cannot stall the airplane and make it go too slow.

[narrator] The A400M is driven by propellers.

That gives it an advantage over traditional jet engines.

The large transport C17,

when it lands on an unprepared strip in the desert,

lots of stones will be sucked up into the engines

and causing engine damage.

and what propellers allow us to do is to land

on unprepared strips without sustaining the type of damage

that a jet engine would suffer.

[narrator] Tony's going to take the A400M

-out for another test drive. -Off we go!

[narrator] But before he does, he's taking us inside,

for an exclusive look at the cockpit,

one of the most advanced in a transport aircraft anywhere.

We've got eight TV screens in the cockpit.

We can configure these screens for each mission

For example, on the displays behind me,

on one display right now we can have the map,

I can put up a terrain display, so I cannot fly into a hill.

Just by pressing one button here

the weather radar is now displayed, which will show

any bad weather ahead of me,

which I can avoid and not fly into.

[narrator] If all that fails, the pilot can still fly

with the head's up display, which shows him

all the essentials. Air speed, altitude,

flight path, and an infrared mode

so he can see at night, even in bad weather.

What it allows the pilot to do is to fly tactically,

for example at low altitude, whilst looking

out of the window. But at the same time,

he's able to read the information

directly in front of his eyes.

[narrator] All the technology provides some comfort.

But there are more comforts built in. It's kind of like

driving an incredibly big, high tech camper van.

It's very comfortable, the seats are well-designed,

the whole space is very well-designed,

and it's a nice place to go to work.

In fact, we even have some very nice and very expensive

meal tray holders where we can have dinner during the flight.

[narrator] No time for lunch. He's gotta fly.

Tony and his colleagues are checking everything

from the plane's maneuverability

to wing performance. The wings are ultralight,

made entirely of carbon fiber.

One of the technologies that stands out

is the use of carbon fiber.

The wings are designed and built with carbon fiber,

as well as the vertical and horizontal stabilizers,

air frame and engine.

This reduces weight and fuel needs.

[narrator] The team is also pushing the limits

of the onboard technologies.

Every flight test, and all data collected in-flight,

is monitored by engineers on the ground

in Spain, France, Germany and England.

[English voice-over] Engineers both in the plane

and on the ground can detect

any flaws in any of the systems

and communicate with the pilots in real-time.

[narrator] Mission accomplished.

[Tony] When the flight test goes right, you definitely feel

a sense of achievement, and you come away feeling

you've done a good day of work.

So clearly that mega-plane is a perfect example

of a man-made machine doing work that we, mere mortals,

can't do alone!

Let's face it, we rely more and more on machines

and computers these days.

But sometimes the simplest human task is not all that

easy for high tech machines.

Now, if you need any proof of that, enter exhibit A.

From the tech team at UC Berkeley,

I present to you, the Quadrotor.

[upbeat music]

[narrator] Most of us take catching a ball for granted.

We see it, track it, we run and then we catch it.

It sounds really easy, right?

But for the computerized Quadrotor helicopter,

It's actually got complicated calculations to make

in order to catch the ball.

Ready... set... go!

The motion capture system finds these reflective dots,

and using this, determines the position

of the Quadrotor helicopter, and to track the ball

we cover it with reflective tape.

[narrator] Humans don't track an airborne ball the same way.

A human baseball player is trying to maintain

a constant angle in their field of view to the ball.

So they're basically tracking it in their visual system,

and trying to keep it roughly in the same spot.

And if that holds the entire time,

eventually the ball has to come towards them.

[narrator] The Quadrotor has got the help

of near-infrared motion capture cameras.

That information helps it predict where the ball

will land, so it can move there to catch it.

At each instant in time, the helicopter is getting

a measurement of the instant position of the ball.

It then uses an idea of the physics of the ball

to predict where it will land.

[narrator] Pat Bouffard got the Quadrotor up and running.

It's quite capable of moving around on its own,

based on commands.

Anil Aswani is now on board, adapting a new technique

he calls "learning-based model predictive control." Wow.

It basically gives machines that were previously automated

the ability to learn, adapt, and make decisions.

The technique is called "predictive control" because

the machine uses its internal model of its behavior

to predict how it's going to behave.

[narrator] In this case, the Quadrotor makes

its predictions based on information it receives

from the cameras set up around the room.

The camera system is kind of a general positioning device

that allows us to figure out where the Quadrotor is

and where the ball is.

[narrator] Instantaneous measurements are happening

one hundred and twenty times a second!

We use the camera system to determine the position

of the ball, and then we look at the position over time

to try to determine what the trajectory of the ball

is gonna be. In other words,

how the ball is gonna go up and curve and come down.

[narrator] The red cross marks the predicted landing spot

of the ball. The Quadrotor constantly adjusts that spot,

throughout the ball's arc.

Because all this happens very quickly, we have to have

a pretty good prediction right off the bat.

[narrator] They need to accurately predict

the ball's trajectory through the air, once it's thrown.

There was a lot of trial and error.

And that was really trying to determine which forces

are the most significant forces on the ball.

Gravity, in this case, is by far the most important factor,

because it's always there, and it's very much constant

and quite predictable.

There's also air resistance,

which tends to slow the ball down,

and that's a fairly well-understood effect.

[narrator] The instant the ball is thrown,

the Quadrotor's prediction is accurate

to within ten centimeters.

But within one second of calculations,

and by the time the ball's about to land,

it's accurate to under one centimeter.

It catches the ball about 90% of the time.

Better than my fielding percentage.

But they still think they can do better.

There's a phenomenon that causes the path of the ball

to curve, because of the spin.

But it ended up below our threshold of modeling,

so we decided not to include that.

But if we were to include this in future,

that might improve our performance and accuracy.

[narrator] So while humans continue to keep

their eye on the ball using their own built-in visual clues,

Anil and Pat will continue to work with the Quadrotor,

which takes its cues from tracking systems

in the outside world.

[Anil] Humans often use approximate techniques that work

very well in practice, whereas machines have to use

very complex mathematical algorithms in order

to replicate what humans do.

And that's a very interesting area of research

on trying to bridge the gap, and that's one

that we may look into in the future

Coming up on How Tech Works.

Here's a car you have to see to believe.

and that's all you have to do,

because this car drives itself... down the Autobahn!

And, we'll check out the world's biggest vertical wind tunnel,

and go sky-diving... indoors!

Welcome back to How Tech Works. I'm Basil Singer,

your host with the most ridiculous love

for all things robot and remote controlled.

So, without further ado,

we catch up with intrepid reporter Dan Riskin,

driving on the super fast Autobahn in Germany,

where, ordinarily, it's critical to stay focused

on the task at hand.

Except when you don't need your hands!

[narrator] The Autobahn. When you drive it,

you've gotta keep your eyes on the road

and your hands upon the wheel.

Unless you're in this car. Look, Ma, no hands!

It's a robot car called Made in Germany,

"MIG" for short.

Made by researchers at the Free University of Berlin.

[speaking German]

[narrator] And today, for the very first time,

They'll let it overtake another car

in heavy Autobahn traffic, all by itself.

But first, a warm up. MIG's going to drive itself

through Berlin traffic.

And I'm going to catch a ride.

[narrator] Two researchers are on board to ensure safety.

But I'm still nervous about this.

Alright, here we are. What are you looking at?

I am looking at the sensor readings of the car.

I have the plan where the car wants to go.

[Dan] So the map of the roads

is from a map that it has in its brain?

[man] Yeah, it's stored in a file.

Right now, I'm in control

and as soon as I put my foot off the brake pedal.

[Dan] Holy shoot! Okay, I can't swear.

[man] So now the car wants to go left on the trajectory.

Three, two, one, go.

So this is driving by itself now?

Yes. It is driving totally by itself.

I don't do anything. My hands are off.

It's going a little fast, it really accelerated quickly.

[man] Yeah it's 50. We didn't want to drive like Grandma,

so it's just normal driving.

[Dan laughs] You didn't want to drive like Grandma?

Is that what you just said? [laughs]

[narrator] The goal is to make an autopilot that can

stop accidents from happening,

and eventually make cars that drive themselves all the time.

Could you go to sleep?

I cannot go to sleep because all the time

there can something happen, and there can be

unexpected situations the car cannot handle,

and then I have to intervene. I only have to put my foot

on the brake or on the gas and then I am in control again.

[Dan] So how long have you had permission to drive this

on the roads of Berlin?

[man] So we just got the permission this week.

This week! So this still pretty brand new for you guys.

Yeah, it is.

[narrator] The team has permission to let MIG

drive itself, back and forth, on this busy boulevard.

[Dan] There's a lot of traffic here!

[man] Yeah, because it's a big road.

[narrator] I try hard to look calm at the first U-turn.

Okay, so now you have a cyclist who is trying

to cross, so this is a spot where she might just pull out.

Yeah, she's going. So the car sees her, and hits the break.

The car recognized her. And she stopped.

She is gone. So we can continue.

[Dan] So that was all the car?

-[man] Yes, it was all the car. -[Dan] That's great!

[narrator] MIG sees the world with a laser sensor up top.

And laser and radar sensors built in the body, and cameras.

The images are processed by a computer in the trunk.

GPS tells it where it is.

Everything feeds into another computer,

which controls motors working the accelerator, brakes,

and steering. Miao's laptop controls it all.

The red shapes are cars and other obstacles.

So we're coming up to a turn there, is that what that shows?

Yeah, it is making a U-turn here.

[narrator] But this is still a work in progress.

The car stops, and starts, and stops.

And when the road is clear, he will pursue his path.

So right now it doesn't start on its own

because there is grass on the street on our right side.

Just this little grass on the side here?

Yeah, so therefore I took over.

And now I give control back to the car.

And now it starts accelerating.

[Dan] That's what that beeping noise was?

[man] Yeah, the beeping noise is always when I intervene.

[narrator] The scanners are picking up overgrown weeds.

The car has some problems at the intersection here.

[narrator] MIG thinks they're obstacles,

so it brakes for them.

Aside from that, they think MIG is performing pretty well.

[car horn beeps]

-Hey, we're autonomous, buddy! -[laughs]

[narrator] Ah, well enough for today's big test.

[Dan] Let's take it up a notch.

We are going to do autonomous driving, on the Autobahn.

[researcher] Okay, Miao.

-Ready. -Oh God! Okay, there.

-That's autonomous now? -Yeah, now it is autonomous.

Okay I take my foot off and now the car takes over.

Whoa!

And now the car follows the car in front of us

so I don't do anything.

So it knows where that car in front of us is

and it just follows it?

[man] Yeah, it follows everything on its lanes.

So you guys are totally calm right now?

Well, we are alert, but we check

if other cars Might cut our lanes.

And how's everything look on the computer?

It's good. All the sensors are up and running.

We drive pretty smoothly on Autobahn.

It checks the velocity of the car in front of us,

and the distance, and from these two parameters

it calculates the speed it wants to go.

[narrator] We're traveling at today's specified top speed.

Maybe you can try an overtaking maneuver now.

So maybe you try you can try an overtaking maneuver,

but I think there are a lot of cars behind us.

Maybe you could slow down to 90.

[narrator] Now I'm really worried.

They want MIG to overtake the car in front of us.

It's driven by one of their colleagues.

When he slows down, MIG will decide

whether it's safe to pass him, or not.

That's what scares me.

Cars come up very quickly in the passing lane

It's not safe, there are a lot of cars back there!

And does it ask permission, or does it just do it?

-It just does it. -Oh my goodness!

[narrator] The car in front slows.

MIG finds a gap in the traffic,

and goes for it.

Daniel intervenes because he can't see the gap.

No, this is too dangerous.

I mean, they come at a hundred kilometers.

[narrator] They try again.

Okay, we are in autonomous mode again.

[indistinct voice on radio]

[narrator] The car in front slows. MIG hits the gas.

Okay, now we overtake.

-[Dan] Wow, that's real. -[man] Yeah.

Wow!

[narrator] MIG passes the passing test.

Not once, but several times.

There's still more work to be done.

But it's a start down the road of driverless cars.

That was a bit of a crazy ride! Thanks again guys!

[car horn beeps]

Finally, as if that wasn't thrilling enough!

We're airborne for our final story.

But there's no aircraft in sight.

In fact, we're even staying indoors... skydiving!

In the world's largest vertical wind tunnel!

[rock music]

[man] Most people see it as a bucket list kind of thing.

Everyone wants to do it once in their life.

[narrator] A giant leap of faith...

begins right here.

[crowd cheers]

[narrator] It's skydiving's inside secret.

What you learn in the tunnel you can take to the sky.

[narrator] Paraclete XP,

the biggest vertical wind tunnel on the planet,

in Raeford, North Carolina.

-It's addictive. -It is addictive!

I want to do it again. [laughs]

[narrator] And while families like this one get sucked

into the excitement in this five meter windy playground,

Every time she comes here she has to do it!

there are some other clients who take things

a little more seriously.

The fact that you can get in the tunnel,

for two, even four minutes at a time, come back out,

watch exactly what you did, debrief it,

and get right back in.

It's unparalleled to be able to actually jump out of a plane,

pack, repack again, go back over it,

and the exhaustion factor comes in to play as well,

so this is an awesome opportunity to have this

local here to Fort Bragg.

[narrator] These are the Golden Knights,

the US Army's high-flying goodwill ambassadors.

[together] Out the hole, into the black.

Once you're there, you never go back. Boom! The Golden Knights.

Drafted from military ranks,

they are like the parachuting elite.

Today, the Golden Knights are training.

With the sound of the bell, they dive right in.

Today we're just going to be working on making points

and being able to fly your body in, next to somebody.

[narrator] Not an easy feat.

The wind speed in here can simulate skydiving

at over 180 miles per hour.

Every flight can be customized in this control room.

Two-thousand horsepower fans

push hot air out past exterior louvers.

New air flows down and into the chamber, where it is spun

around by turn levers, making it virtually turbulence free.

You realize little things make a huge difference,

like your hand corking a little bit one way,

or your foot turning out another,

and our boss can get in there, tell us what we're doing

wrong or right, and we execute it and make a difference.

[narrator] Much better to make mistakes in here

than in the skies.

Still, it's dangerous.

A collision at this speed could cause serious injury.

Sometimes you have your day, sometimes you don't.

It's like any other sport in the world.

[narrator] A sport that sparks wonder at airshows

and inspires others to fly.

[man] There's nothing like it in the world.

Absolutely nothing.

People ride planes every day, but how many people

can get out of a plane and just fall all on their own?

It's awesome.

[narrator] Awesome enough that this family

is getting a jump start.

Anybody can do it. Whether you're a kid, or 60,

or a grandparent, a mom, a dad, doesn't matter.

So it's really cool.

Wow, what a great way of experiencing

non-stop sky diving without the fear of jumping

out of a plane.

Well, that's all we've got time for today.

Thank you very much for watching How Tech Works.

Until next time, me, Dr. Basil Singer!

[ending music]