Dawn of the Driverless Car

Download Subtitles

Transcript

0:00:03 > 0:00:06The motorcar has shrunk the world,

0:00:06 > 0:00:08increased personal freedom,

0:00:08 > 0:00:11and in so many ways, expanded our horizons.

0:00:12 > 0:00:14But there's a flip side.

0:00:14 > 0:00:16Cars have destroyed our environment,

0:00:16 > 0:00:18poisoned the air we breathe,

0:00:18 > 0:00:22and killed us in far more straightforward ways.

0:00:23 > 0:00:27There's about a million deaths every year in this world due to traffic

0:00:27 > 0:00:29accidents and I find this just

0:00:29 > 0:00:32utterly unacceptable in the 21st century.

0:00:32 > 0:00:35But all that's going to change.

0:00:35 > 0:00:39Soon, we'll be in a position to have our automotive cake and eat it.

0:00:40 > 0:00:44Self-driving cars are going to have a huge impact on society.

0:00:44 > 0:00:47They'll be able to navigate through complex intersections

0:00:47 > 0:00:48with no collisions.

0:00:48 > 0:00:51If I could just sit back and read a book, listen to music,

0:00:51 > 0:00:53catch up on some sleep, that would be great.

0:00:53 > 0:00:56Once your car can come from round the corner,

0:00:56 > 0:00:59we can start opening up some of these sort of residential areas and

0:00:59 > 0:01:01they'll feel much more sociable again.

0:01:01 > 0:01:05This is a world where cars will drive themselves,

0:01:05 > 0:01:07a world where we are simply passengers,

0:01:07 > 0:01:09ferried about by wholesome, green,

0:01:09 > 0:01:13compassionate technology which will never, ever go wrong.

0:01:13 > 0:01:15And it's almost here.

0:01:15 > 0:01:17I can press this button.

0:01:17 > 0:01:20I don't even have a control that I can grab.

0:01:20 > 0:01:26A fully autonomous vehicle in commercial operation in 2021.

0:01:27 > 0:01:31But cars that can run errands for us by themselves could quickly clog up

0:01:31 > 0:01:34our streets and ruin livelihoods, too.

0:01:36 > 0:01:39From taxi drivers to truck drivers,

0:01:39 > 0:01:43lots of people have a job as a driver.

0:01:43 > 0:01:48A lot of effort has been put into selling us the driverless dream.

0:01:48 > 0:01:50Now it's almost upon us,

0:01:50 > 0:01:53could we actually be sleepwalking into a nightmare?

0:01:54 > 0:01:57What happens when it eventually encounters a no-win scenario,

0:01:57 > 0:01:59when it actually has to have an accident?

0:02:13 > 0:02:14We love cars.

0:02:14 > 0:02:17We love owning them,

0:02:17 > 0:02:20we love driving them,

0:02:20 > 0:02:23and learning to drive a car is a rite of passage.

0:02:23 > 0:02:25Handbrake off. Handbrake.

0:02:25 > 0:02:27- Oop.- Handbrake.

0:02:27 > 0:02:30You've got to be able to drive or else you're really not a real adult.

0:02:32 > 0:02:37But learning to mirror, signal and manoeuvre is on its way out,

0:02:37 > 0:02:40because shadowy backroom technologists at places like Google,

0:02:40 > 0:02:42Intel and even Facebook

0:02:42 > 0:02:45are hell-bent on getting rid of drivers altogether.

0:02:46 > 0:02:50We're building systems to enable cars to drive themselves.

0:02:54 > 0:02:57Our lovely cars are on borrowed time.

0:02:57 > 0:03:01There's even a plan drawn up by the Society of Automotive Engineers.

0:03:01 > 0:03:03A road map, if you will.

0:03:04 > 0:03:07It starts with level zero.

0:03:07 > 0:03:11Level zero is where the human driver has control of everything.

0:03:12 > 0:03:14The plan then moves through various

0:03:14 > 0:03:17levels of automation and driver-assist

0:03:17 > 0:03:20technologies and ends up at level five.

0:03:21 > 0:03:24Level five is where humans are simply passengers.

0:03:25 > 0:03:28At this point, the word "driving"

0:03:28 > 0:03:31reverts to being something to do with livestock.

0:03:33 > 0:03:36If this sounds a bit like a science fiction writer's pipe dream,

0:03:36 > 0:03:40then remember, an awful lot of effort and, crucially, money

0:03:40 > 0:03:42is currently going into it.

0:03:43 > 0:03:47Google is investing 30 million annually to driverless,

0:03:47 > 0:03:50and Intel recently paid 15 billion

0:03:50 > 0:03:53for Israeli driverless tech company Mobileye.

0:03:55 > 0:03:58From niche robot race cars...

0:03:58 > 0:04:00Look, it's being steered by nobody!

0:04:01 > 0:04:04..to mass manufacturers,

0:04:04 > 0:04:06driverless is where it's at.

0:04:07 > 0:04:12I think the promise of a driverless future where cars are available to

0:04:12 > 0:04:14everyone and they can do all the

0:04:14 > 0:04:19hard work and ease congestion and ease pollution is very exciting,

0:04:19 > 0:04:20but I am a bit doubtful as to

0:04:20 > 0:04:23whether that's how it will actually play out.

0:04:23 > 0:04:25Autonomy is probably the biggest

0:04:25 > 0:04:28thing that's being talked about in the automotive industry.

0:04:28 > 0:04:31People that are less able will be able to get around because they

0:04:31 > 0:04:33won't need to be driving. But obviously,

0:04:33 > 0:04:36there's so much legislation to get in place to get to the point where

0:04:36 > 0:04:39- this is actually viable.- I do think people want...

0:04:39 > 0:04:42In principle, they want to be able to move around with less effort.

0:04:42 > 0:04:44I think everybody wants that.

0:04:44 > 0:04:46I'm not sure that's what Ford is talking about when it says it's

0:04:46 > 0:04:49going to have a driverless car by 2020, or whatever it is.

0:04:50 > 0:04:52It sounds exciting,

0:04:52 > 0:04:55but I think it's...

0:04:55 > 0:04:58I want to make sure it works, and that's the biggest challenge.

0:04:58 > 0:05:01Is it going to do what it's supposed to do?

0:05:01 > 0:05:03Will I trust it?

0:05:03 > 0:05:06Whatever eventually emerges onto the roads of tomorrow,

0:05:06 > 0:05:09the future looks bleak for the intimate relationship we currently

0:05:09 > 0:05:12enjoy with our fine, four-fendered friends.

0:05:20 > 0:05:24This is a momentous day in Harrison's life.

0:05:24 > 0:05:27Today he is having his first driving lesson.

0:05:28 > 0:05:31I learned to drive when I was 17.

0:05:31 > 0:05:33I was quite nervous.

0:05:33 > 0:05:34I loved learning to drive. I'd been

0:05:34 > 0:05:36looking forward to it for quite a long time.

0:05:36 > 0:05:38I was failed pretty much about

0:05:38 > 0:05:4015 minutes in for failing to give way to someone.

0:05:40 > 0:05:43Harrison has never attempted to drive before,

0:05:43 > 0:05:44and now he's nailed the

0:05:44 > 0:05:47all-important selfie with his new best friend,

0:05:47 > 0:05:49there's a lot to think about.

0:05:49 > 0:05:50Push that all the way down?

0:05:50 > 0:05:53Yeah, press the brake all the way down and press the start button.

0:05:53 > 0:05:55Look all around the car like this.

0:05:55 > 0:05:57Just pop your signal on, press the button in,

0:05:57 > 0:05:59pull it up and ease up off your clutch, slowly.

0:05:59 > 0:06:02- That's it.- OK.- And we're just going to go off like that.

0:06:02 > 0:06:04- OK.- Have you got the idea? - Yeah.

0:06:04 > 0:06:06In a few short minutes, however,

0:06:06 > 0:06:11Harrison is transformed from ordinary mortal to driver.

0:06:11 > 0:06:13Exciting, isn't it? Yo, let's go!

0:06:13 > 0:06:16- Let's go.- Don't worry, don't worry.

0:06:16 > 0:06:18- That's fine.- OK, so he's not the finished article,

0:06:18 > 0:06:20but in a couple of months,

0:06:20 > 0:06:22he'll hopefully have passed his driving test and

0:06:22 > 0:06:26will think nothing more of driving than he currently does of walking.

0:06:26 > 0:06:28There you go. You're off.

0:06:28 > 0:06:31You'll be driving back home in no time.

0:06:31 > 0:06:32In simple terms,

0:06:32 > 0:06:36Harrison has sensed the world around him and reacted appropriately.

0:06:36 > 0:06:38It's what he's been doing all his life.

0:06:39 > 0:06:45- Well done.- Only today, he's learning to do that via a machine.

0:06:45 > 0:06:49Sensing our environment is something most of us take wholly for granted.

0:06:49 > 0:06:53It's part of being human, something we're good at.

0:06:53 > 0:06:56But little by little, cars have been getting in on the act, too.

0:07:03 > 0:07:08Parking sensors, lane sensors, cruise control,

0:07:08 > 0:07:10adaptive cruise control,

0:07:10 > 0:07:14automatic headlights and windscreen wipers have all emerged under the

0:07:14 > 0:07:16banner of driver-assist technologies.

0:07:18 > 0:07:21But in reality, they are the first,

0:07:21 > 0:07:25small steps on the road to full autonomy, and that's official.

0:07:25 > 0:07:29It's the first stage on the driverless masterplan.

0:07:31 > 0:07:34Level one.

0:07:34 > 0:07:38At level one, the vehicle can take control of individual functions like

0:07:38 > 0:07:40acceleration or braking.

0:07:41 > 0:07:44Effectively, cars can actually drive

0:07:44 > 0:07:46themselves to a small degree already.

0:07:47 > 0:07:52Electronic stability control or ABS is one of the systems that's been

0:07:52 > 0:07:55put in that stops me from skidding and crashing.

0:07:55 > 0:07:58Adaptive cruise control brakes and speeds up for you,

0:07:58 > 0:08:00depending on what the car in front

0:08:00 > 0:08:02is doing, and that is just so great for long journeys.

0:08:02 > 0:08:05Like, I drove to Wales on the M4 the other week and it's a

0:08:05 > 0:08:09really long journey and it was just effortless because of that.

0:08:09 > 0:08:11Many of these automotive technologies

0:08:11 > 0:08:13have their origins at the racetrack.

0:08:17 > 0:08:20And even though car racing is about as driver-focused

0:08:20 > 0:08:22as it's possible to be...

0:08:24 > 0:08:26..there is some common ground

0:08:26 > 0:08:28between Formula 1 and the driverless future.

0:08:32 > 0:08:35My name is Teena Gade, and I work here at Sahara Force India Formula 1

0:08:35 > 0:08:37team as a vehicle science engineer.

0:08:38 > 0:08:41The best place normally to develop a car is actually to take it to a

0:08:41 > 0:08:42track and test it. In reality,

0:08:42 > 0:08:45we're restricted by the regulations so we're only allowed a fixed number

0:08:45 > 0:08:47of test sessions every year.

0:08:47 > 0:08:50What that means is we end up having to do quite a lot of it in the

0:08:50 > 0:08:52virtual world. We model the car, the tyres,

0:08:52 > 0:08:54the aerodynamics and the tracks and

0:08:54 > 0:08:56we bring it all together in a driving

0:08:56 > 0:08:58simulator for a driver to drive around and it tells us what the

0:08:58 > 0:09:00performance is going to be.

0:09:00 > 0:09:03One of the most interesting things for me about the concept of

0:09:03 > 0:09:05driverless cars is if you take what I do on a daily basis

0:09:05 > 0:09:07but we really don't understand,

0:09:07 > 0:09:10the bit that is difficult is the driver element, because for all the

0:09:10 > 0:09:13computational power we have to model, for example,

0:09:13 > 0:09:15the aerodynamics or the tyres or the track,

0:09:15 > 0:09:18the things that go on in the human brain are incredibly complicated and

0:09:18 > 0:09:20processed unbelievably quickly.

0:09:20 > 0:09:23So from an engineering perspective,

0:09:23 > 0:09:25if we could have a completely driverless car it could perform the

0:09:25 > 0:09:29same task absolutely consistently, the same from one lap to the next,

0:09:29 > 0:09:32every time. We would get much cleaner data and actually,

0:09:32 > 0:09:34we potentially stand to make better connections

0:09:34 > 0:09:36as a result and learn more.

0:09:38 > 0:09:41Today, Teena is off to see another race team,

0:09:41 > 0:09:45one that have taken driverless technology a stage further.

0:09:46 > 0:09:48I actually think I want to go left here.

0:09:48 > 0:09:51No, I was completely wrong. I can go that way.

0:09:54 > 0:09:58Roborace want to compete in Formula E,

0:09:58 > 0:10:00but unlike other electric race teams,

0:10:00 > 0:10:03they have decided to abandon drivers altogether.

0:10:05 > 0:10:09These sleek racing robots will battle it out against each other and

0:10:09 > 0:10:11eventually against human competition.

0:10:12 > 0:10:16But for the time being, they're testing out the concept with this,

0:10:16 > 0:10:19their development robot, DevBot.

0:10:21 > 0:10:23We can see inside here,

0:10:23 > 0:10:26it actually looks quite a lot like a conventional car cabin.

0:10:26 > 0:10:29Yeah, we've got a rack that sits behind the driver.

0:10:29 > 0:10:32What we've done is we've taken human capabilities and we're putting

0:10:32 > 0:10:34them in silicon and software.

0:10:36 > 0:10:39Here at Silverstone, DevBot knows

0:10:39 > 0:10:41the circuit so well that it's capable of

0:10:41 > 0:10:45giving high-speed tours of the racetrack to the humans it's aiming

0:10:45 > 0:10:49- to defeat.- I'm not actually the best passenger at the best of times,

0:10:49 > 0:10:52and so it's going to be quite strange.

0:10:52 > 0:10:54Hold the handbrake.

0:10:54 > 0:10:56- I would drive like that.- OK.

0:10:58 > 0:11:00So this is it.

0:11:00 > 0:11:02My life in the hands of some software.

0:11:02 > 0:11:04The blue light's gone on at the front.

0:11:04 > 0:11:05I think we're ready to go.

0:11:05 > 0:11:07I am really very nervous.

0:11:14 > 0:11:18It's terrifying, the first time you get in a car and you're not touching

0:11:18 > 0:11:19any of the controls.

0:11:39 > 0:11:40Once you've spent some time in it,

0:11:40 > 0:11:43you get that feeling of total confidence,

0:11:43 > 0:11:44and you realise that the machine is

0:11:44 > 0:11:46far better than any human could ever be.

0:11:46 > 0:11:48Oh, wow!

0:11:48 > 0:11:50That is hard on the brakes.

0:11:50 > 0:11:53Can I have another go? That's actually quite good!

0:11:53 > 0:11:56One of the big appeals about Formula 1 and motorsport in general is the

0:11:56 > 0:11:59personalities, the drivers themselves.

0:11:59 > 0:12:02There is also a huge following for teams.

0:12:02 > 0:12:04It's quite good in there, it's quite good, yeah!

0:12:04 > 0:12:07If you look at Ferrari, they're one of the biggest in the world.

0:12:07 > 0:12:10And I think what Roborace will allow is actually if people create that

0:12:10 > 0:12:14following for the team rather than for the driver themselves.

0:12:15 > 0:12:18Impressive as careering driverless

0:12:18 > 0:12:21around the track at around 200kph is,

0:12:21 > 0:12:25there's much more to racing than just pure speed.

0:12:25 > 0:12:28A driverless race car will have a lot on its mind.

0:12:30 > 0:12:33What we're really starting to look at is that judgment layer.

0:12:33 > 0:12:36"What speed should I be entering this corner?" -

0:12:36 > 0:12:37is the critical thing.

0:12:37 > 0:12:40The path that I should be following.

0:12:40 > 0:12:43And then moving up into sort of the tactical decision-making layer in

0:12:43 > 0:12:46terms of - "Am I going to overtake or am I going to actually save some

0:12:46 > 0:12:50"battery because I want to attack in about five laps' time?"

0:12:50 > 0:12:51So what you're basically saying is

0:12:51 > 0:12:55that actually the car can perform the driving function as a human can?

0:12:55 > 0:12:57Yes, that's exactly right, yeah.

0:12:58 > 0:13:01It is Roborace's plan to build such

0:13:01 > 0:13:03a clever driverless car that the human

0:13:03 > 0:13:06opposition will be ground into the dust.

0:13:06 > 0:13:09But so far, the DevBots are only racing each other.

0:13:09 > 0:13:11With mixed results.

0:13:14 > 0:13:16And this is the problem.

0:13:16 > 0:13:19Robots, even quite advanced ones

0:13:19 > 0:13:24like the ones that run DevBot, are, well, limited.

0:13:28 > 0:13:30Here is a representative selection.

0:13:32 > 0:13:34Looking at these brand ambassadors,

0:13:34 > 0:13:37the future for autonomous cars doesn't look over-rosy.

0:13:38 > 0:13:42If they stand any chance of gaining our trust,

0:13:42 > 0:13:45they're going to have to deliver a lot more than these.

0:13:46 > 0:13:48And DevBot.

0:13:53 > 0:13:56Yeah, so we're turning left. Ease up very slowly off the clutch.

0:13:56 > 0:13:59And start to steer towards me.

0:13:59 > 0:14:01To successfully avoid crashing his car,

0:14:01 > 0:14:05Harrison will have to be able to perform a bewildering array of tasks

0:14:05 > 0:14:08simultaneously in real time.

0:14:08 > 0:14:10You've gone in the wrong gear, don't worry.

0:14:10 > 0:14:12- Oh, sorry.- I've helped you out. Don't worry.

0:14:12 > 0:14:14That's part of being a learner.

0:14:14 > 0:14:17- OK.- Harrison will have to be able to recognise, categorise,

0:14:17 > 0:14:19and accurately predict the likely

0:14:19 > 0:14:22future actions of anything he sees while driving.

0:14:22 > 0:14:24Check this mirror

0:14:24 > 0:14:27and put your left signal on, which is downwards.

0:14:27 > 0:14:28In the light of those

0:14:28 > 0:14:31near instantaneously acquired pieces of information,

0:14:31 > 0:14:34he must cause the car to safely accelerate, change direction,

0:14:34 > 0:14:38slow down or stop, or all or none of the above.

0:14:38 > 0:14:40Hang on, I'm just going to stop you there.

0:14:40 > 0:14:42We just had a car coming rather quickly.

0:14:46 > 0:14:51And bear in mind that this is a constantly updated stream of data to

0:14:51 > 0:14:54be interpreted and acted upon by Harrison,

0:14:54 > 0:14:55millisecond by millisecond,

0:14:55 > 0:14:58for as long as he's behind the wheel.

0:14:58 > 0:15:01We just had a narrow escape with that little caravan there.

0:15:02 > 0:15:07That, in a nutshell, is what will be required of a self-driving car.

0:15:07 > 0:15:12It too will need to understand and interact with its environment.

0:15:12 > 0:15:14But let's not get ahead of ourselves.

0:15:14 > 0:15:17- We're not going down that busy one, don't worry.- Yeah, good!

0:15:18 > 0:15:21We're only just getting to level two, after all.

0:15:21 > 0:15:23Level two.

0:15:23 > 0:15:27Level two is the point on the road to autonomy at which the would-be

0:15:27 > 0:15:31driverless car can control two things at the same time,

0:15:31 > 0:15:33like steering AND braking.

0:15:35 > 0:15:38It's a small but significant step.

0:15:38 > 0:15:41But before we send all our current cars to the crusher,

0:15:41 > 0:15:45it's worth remembering that some of them can multitask already.

0:15:46 > 0:15:50Automatic parking I think people are very happy to use.

0:15:50 > 0:15:52It scans for a car parking space,

0:15:52 > 0:15:54and then you just have to do the

0:15:54 > 0:15:56gas and brake while it does the steering for you.

0:15:56 > 0:15:59All these tricks are all well and good,

0:15:59 > 0:16:01and of course really very clever.

0:16:01 > 0:16:03But to be fair, they're not exactly

0:16:03 > 0:16:06difficult for human drivers to pull off.

0:16:06 > 0:16:08It makes you wonder why we're

0:16:08 > 0:16:10bothering with driverless cars at all.

0:16:10 > 0:16:12The main attractive feature of

0:16:12 > 0:16:17driverless cars is the promise that they will save lives.

0:16:17 > 0:16:19I don't think humans are that great at driving.

0:16:19 > 0:16:22You only have to look at the statistics to see that

0:16:22 > 0:16:25there is a need for improvement there.

0:16:25 > 0:16:30So if automating the driver can reduce those deaths,

0:16:30 > 0:16:32then it's definitely a desirable thing.

0:16:33 > 0:16:36There are systems that can detect whether you fall asleep or not.

0:16:36 > 0:16:40And if you don't wake up, then the car will perform an emergency stop.

0:16:40 > 0:16:43And some really cool stuff has been happening in collision avoidance,

0:16:43 > 0:16:45so the car taking over when it

0:16:45 > 0:16:48thinks you're going to have an accident

0:16:48 > 0:16:50and you haven't reacted quickly enough.

0:16:50 > 0:16:54The number of accidents that's saved alone over the last couple of years

0:16:54 > 0:16:55is phenomenal.

0:16:57 > 0:17:01The promise of a safer-than-human driverless car has been around for

0:17:01 > 0:17:03almost as long as the car itself.

0:17:07 > 0:17:10We trust our lives in machines all the time.

0:17:12 > 0:17:14Every time you get on a commercial flight,

0:17:14 > 0:17:16you're not being flown by a human any more.

0:17:16 > 0:17:18And that's good, because you're safer this way.

0:17:18 > 0:17:20Sebastian Thrun has spent most of

0:17:20 > 0:17:23his professional life trying to bring

0:17:23 > 0:17:25that airline level of safety to our cars.

0:17:26 > 0:17:30I got involved because I had a traumatic event as a teenager.

0:17:30 > 0:17:31My best friend died in a traffic

0:17:31 > 0:17:34accident from one moment to the next.

0:17:34 > 0:17:37And I found his death kind of a little bit ridiculous.

0:17:37 > 0:17:38I think we don't talk about that much,

0:17:38 > 0:17:41but there's about 1 million or 1.2 million

0:17:41 > 0:17:44deaths every year in this world to traffic accidents.

0:17:44 > 0:17:47And I find this just utterly unacceptable in the 21st century.

0:17:47 > 0:17:49So I wanted to fix that.

0:17:49 > 0:17:52Given that 90% of traffic accidents are due to

0:17:52 > 0:17:56human error, Sebastian decided that the best and safest course of action

0:17:56 > 0:17:59would be to dispense with the driver altogether.

0:18:03 > 0:18:05In 2005, he had a breakthrough.

0:18:07 > 0:18:11My team at Stanford built the car that won the DARPA Grand Challenge,

0:18:11 > 0:18:14a desert race for a car that could drive itself.

0:18:16 > 0:18:20The car crossed more than 200 kilometres of desert in a little

0:18:20 > 0:18:24over six hours, with no human driver and no human intervention.

0:18:26 > 0:18:28Sebastian was delighted.

0:18:32 > 0:18:37The DARPA Grand Challenge winner and DevBot show a glimpse of the future.

0:18:37 > 0:18:40But the trouble is that the world of driving doesn't usually exist on a

0:18:40 > 0:18:44racetrack of known shape, size and camber

0:18:44 > 0:18:46or in a relatively benign desert,

0:18:46 > 0:18:48where the worst that could happen is cactus damage.

0:18:50 > 0:18:54Real driving happens in the temporarily reversed Croydon one-way

0:18:54 > 0:18:56system on a wet Thursday morning.

0:19:00 > 0:19:05Formula 1 engineer Teena Gade has been impressed by the idea of robot

0:19:05 > 0:19:07race cars. But like the rest of us,

0:19:07 > 0:19:08she is keen to discover how

0:19:08 > 0:19:13driverless cars might help with the real world.

0:19:13 > 0:19:15Driving is pleasurable, say, for example,

0:19:15 > 0:19:17you're driving on the west coast of California.

0:19:17 > 0:19:19But if you do it for ten hours a week,

0:19:19 > 0:19:22suddenly it's not that much fun any more.

0:19:22 > 0:19:24The reality is that if there was a

0:19:24 > 0:19:26train that went between where I lived and worked,

0:19:26 > 0:19:27I'd probably take the train,

0:19:27 > 0:19:30because it would mean I could be doing something more productive.

0:19:30 > 0:19:33So if I had a machine that could do that for me,

0:19:33 > 0:19:35then that would be a great thing.

0:19:35 > 0:19:37That would buy me back ten hours a week.

0:19:37 > 0:19:38That's a whole working day.

0:19:41 > 0:19:44Today, Teena's off to visit a tech start-up company who want to help.

0:19:46 > 0:19:50FIVE AI is the brainchild of Stan Boland,

0:19:50 > 0:19:52who aims to turn this reasonably

0:19:52 > 0:19:56priced electric car into the star of the driverless world.

0:19:57 > 0:19:59So why would you call your company FIVE AI?

0:19:59 > 0:20:01We're aiming for, ultimately,

0:20:01 > 0:20:04the highest level of autonomy, which is level five,

0:20:04 > 0:20:06a car that is utterly autonomous,

0:20:06 > 0:20:10can drive anywhere without any kind of human intervention.

0:20:10 > 0:20:13In fact, it hasn't even got provision for human intervention.

0:20:13 > 0:20:14So we called the company FIVE AI.

0:20:14 > 0:20:16Oh, wow, OK, that makes sense.

0:20:16 > 0:20:19Urban driving creates the most challenges, actually,

0:20:19 > 0:20:21for an autonomous car system.

0:20:21 > 0:20:25You may have cyclists and pedestrians and cars,

0:20:25 > 0:20:27complex buildings,

0:20:27 > 0:20:30road markings that confuse, and people might come from any direction

0:20:30 > 0:20:32and do almost anything in front of you, really.

0:20:34 > 0:20:38Stan and the team are building their system from scratch, and today,

0:20:38 > 0:20:40they're installing the sensors the

0:20:40 > 0:20:43car will need to negotiate the world.

0:20:43 > 0:20:45It needs to be able to see,

0:20:45 > 0:20:48so we use cameras for that and we use light radar.

0:20:48 > 0:20:50We also use ultrasound,

0:20:50 > 0:20:53just like we would in our cars at

0:20:53 > 0:20:55home to detect objects that are very, very close by.

0:20:55 > 0:20:59But when you get to sort of solve problems like fog or

0:20:59 > 0:21:02night-time, or rain or snow,

0:21:02 > 0:21:05then it requires infrared cameras or high-sensitivity cameras,

0:21:05 > 0:21:07so we should be able to sort of

0:21:07 > 0:21:10drive along and just reconstruct the world as we go.

0:21:10 > 0:21:11Just like humans do, really.

0:21:11 > 0:21:14So the vision systems are actually just the tip of the iceberg.

0:21:14 > 0:21:17What you're talking about is going on to then what's happening in the

0:21:17 > 0:21:19- brains of this car?- That's right,

0:21:19 > 0:21:22it does require a huge amount of cognitive capability in the car to

0:21:22 > 0:21:26be able to sort of sense the world and then take decisions about how we

0:21:26 > 0:21:29control it. So that is a huge problem,

0:21:29 > 0:21:33but one that we think we can actually solve.

0:21:33 > 0:21:36To do that, Stan and the team will need to get their system through all

0:21:36 > 0:21:40the levels of autonomy, including level three.

0:21:43 > 0:21:45Level three.

0:21:46 > 0:21:48At level three, the masterplan

0:21:48 > 0:21:51states that safety-critical functions can

0:21:51 > 0:21:54be completely assigned to the vehicle under certain driving or

0:21:54 > 0:21:55environmental conditions...

0:21:56 > 0:21:59..but that a supervising human driver

0:21:59 > 0:22:02must be present to take over in emergencies.

0:22:03 > 0:22:07What that means in reality is that if the driving and environmental

0:22:07 > 0:22:09conditions allow, you can shout at

0:22:09 > 0:22:13the children face-to-face while the car drives itself.

0:22:13 > 0:22:17Now, this is much more like it, and what's more,

0:22:17 > 0:22:19it's almost available now.

0:22:20 > 0:22:23And this is what it might look like in reality.

0:22:23 > 0:22:27Or at least, what it looks like in the mind of whichever creative Volvo

0:22:27 > 0:22:29hired to create this.

0:22:29 > 0:22:32Once the two green bars in the centre meet,

0:22:32 > 0:22:36the paddle lights shift to green and the autopilot confirms that the

0:22:36 > 0:22:39driving and the supervision is delegated to the car.

0:22:42 > 0:22:43And this is what this slightly

0:22:43 > 0:22:46nervous-looking man from Nissan unleashed

0:22:46 > 0:22:50on some carefully selected parts of east London earlier this year.

0:22:51 > 0:22:54This is a real car and these are real streets.

0:22:54 > 0:22:56And those are real hands,

0:22:56 > 0:23:00poised to grab the autonomous steering wheel at a moment's notice.

0:23:01 > 0:23:03Being in an autonomous car is

0:23:03 > 0:23:07obviously a weird feeling because you're doing nothing.

0:23:07 > 0:23:09It's quite unnerving the first time you do it,

0:23:09 > 0:23:12certainly handing over control on a motorway, for example,

0:23:12 > 0:23:14when it's obviously high speed.

0:23:14 > 0:23:17But what's surprising is how quickly you become accustomed to the fact

0:23:17 > 0:23:20that the car is driving itself.

0:23:22 > 0:23:24Driverless cars are kept on the

0:23:24 > 0:23:26straight and narrow using three things.

0:23:26 > 0:23:31First, satellite navigation knows, more or less, where the road is.

0:23:31 > 0:23:33Lidar, a spinning laser,

0:23:33 > 0:23:37builds up an overview picture of the environment by recording its own

0:23:37 > 0:23:40reflection, and radar does a similar thing for the short range.

0:23:42 > 0:23:45It's this combination that Sebastian Thrun and his team used

0:23:45 > 0:23:48to autonomously navigate through the Mojave Desert

0:23:48 > 0:23:50and win the DARPA Challenge.

0:23:52 > 0:23:55But for autonomously driving you home from a drink-and-drugs-fuelled

0:23:55 > 0:23:59weekend with your crazy friends on the coast, it just can't cut it.

0:24:00 > 0:24:02You still need a human, and that's a problem...

0:24:04 > 0:24:07..one that's being addressed at Stanford University in California.

0:24:10 > 0:24:15I'm going to get over this hill and then I'll talk to you.

0:24:15 > 0:24:19I'm Wendy Ju and I work on research particularly around how people are

0:24:19 > 0:24:21going to interact with automation.

0:24:21 > 0:24:25So we're basically borrowing a little piece of the possible future

0:24:25 > 0:24:28and then running experiments to see how people react there.

0:24:28 > 0:24:31So right now, we're in a driving simulator.

0:24:31 > 0:24:33In this particular set-up, we are trusting

0:24:33 > 0:24:35this robot steering wheel.

0:24:35 > 0:24:37I can press this button.

0:24:38 > 0:24:40Now the car is going to switch to autonomy,

0:24:40 > 0:24:43but the steering wheel handle actually goes back.

0:24:43 > 0:24:46So if I'm a little distracted, daydreaming and I look down,

0:24:46 > 0:24:47I know who's in charge,

0:24:47 > 0:24:50not only because there's this little icon on the dashboard,

0:24:50 > 0:24:52but I don't even have a control I can grab.

0:24:53 > 0:24:55The problem for the driver, though,

0:24:55 > 0:24:59is paying attention when the machine is in control.

0:24:59 > 0:25:03Human minds tend to wander, get distracted, or simply shut down.

0:25:04 > 0:25:08The idea of autonomous cars is really, really exciting.

0:25:08 > 0:25:13So it's ironic that the reality of being in an autonomous car is that

0:25:13 > 0:25:16being in an autonomous car is really, really boring.

0:25:16 > 0:25:18It goes against intuition,

0:25:18 > 0:25:22but what our research points to is how to keep people engaged and alert

0:25:22 > 0:25:25and aware of what's going on on the road.

0:25:25 > 0:25:27Because when we run an experiment,

0:25:27 > 0:25:29people who are told just to

0:25:29 > 0:25:31supervise the car are falling asleep.

0:25:31 > 0:25:34Just the task of supervising the car

0:25:34 > 0:25:37is not engaging enough to keep people awake.

0:25:37 > 0:25:40And this is the Achilles heel of level three autonomy.

0:25:40 > 0:25:45There needs to be a human ready to avert disaster at all times.

0:25:45 > 0:25:50So we have this camera here pointed at my face to tell if they're awake

0:25:50 > 0:25:53or asleep, if they're alert or not.

0:25:53 > 0:25:56I think that it's going to become important for the car to do active

0:25:56 > 0:25:58intervention. Maybe talking to people more,

0:25:58 > 0:26:01entertain them more if they look drowsy or asleep.

0:26:01 > 0:26:02Ask them questions, you know,

0:26:02 > 0:26:05"Do you think we should go this way or that way?"

0:26:05 > 0:26:07You know, "What do you think of this thing over there?"

0:26:07 > 0:26:10If only the humans inside had something to do.

0:26:10 > 0:26:11Like steer.

0:26:13 > 0:26:14Just a thought.

0:26:16 > 0:26:20I think people want fully driverless cars that can do absolutely

0:26:20 > 0:26:23everything without human input.

0:26:24 > 0:26:26I'm less convinced that people will

0:26:26 > 0:26:28be happy with a supposedly driverless

0:26:28 > 0:26:30car that actually needs them to be

0:26:30 > 0:26:33paying attention the whole time in case something goes wrong.

0:26:33 > 0:26:36What happens when we get used to not driving for

0:26:36 > 0:26:40long periods of time and then are asked to act quickly if something

0:26:40 > 0:26:42happens and take control,

0:26:42 > 0:26:45and our reactions are not fast enough to actually do that?

0:26:45 > 0:26:49Will they be completely distracted reading a magazine and so not

0:26:49 > 0:26:51realise they have to take control?

0:26:51 > 0:26:53That's one of the biggest challenges.

0:26:54 > 0:26:57Level three automation, then, is slightly pointless,

0:26:57 > 0:27:00a kind of cleverer cruise control

0:27:00 > 0:27:02with a bit of steering sometimes thrown in.

0:27:02 > 0:27:04What we really need is something

0:27:04 > 0:27:07that can take on everything a human driver can.

0:27:07 > 0:27:11Something clever. Intelligent, even.

0:27:14 > 0:27:16Luckily, help is at hand on the

0:27:16 > 0:27:18other side of the San Francisco Bay Area.

0:27:20 > 0:27:24The University of California at Berkeley is Stanford's groovier,

0:27:24 > 0:27:26longer-haired, more radical cousin.

0:27:29 > 0:27:32Scientists here have come up with this.

0:27:32 > 0:27:34It may not look radical,

0:27:34 > 0:27:37but what it can do may just deliver the driverless revolution we've all

0:27:37 > 0:27:38been promised.

0:27:40 > 0:27:44Brett is the Berkeley Robot for the Elimination of Tedious Tasks.

0:27:45 > 0:27:47If you look here, we got Brett's toys.

0:27:47 > 0:27:51Brett has learned to stack Lego blocks,

0:27:51 > 0:27:55has learned to put caps onto bottles,

0:27:55 > 0:27:57assembling pieces of this airplane.

0:27:57 > 0:28:00Whilst this might seem slightly underwhelming,

0:28:00 > 0:28:04the key here is the idea of learning.

0:28:04 > 0:28:08Nobody has written Brett a bottle-top program or a block-stacking

0:28:08 > 0:28:13algorithm. The robot has worked it out for himself,

0:28:13 > 0:28:16something that until recently was the exclusive domain

0:28:16 > 0:28:18of animals with sizeable brains.

0:28:19 > 0:28:21I am Pieter Abbeel.

0:28:21 > 0:28:24I'm a professor at UC Berkeley and I work on artificial intelligence for

0:28:24 > 0:28:26robotics. Basically,

0:28:26 > 0:28:29robots are mechanically very capable and can do a lot of things.

0:28:29 > 0:28:31But in practice, they do very little for us.

0:28:31 > 0:28:34What's holding them back is their lack of intelligence.

0:28:37 > 0:28:39Or at least it was.

0:28:39 > 0:28:43Artificial intelligence, also known as machine learning,

0:28:43 > 0:28:47is allowing robots to acquire new skills for themselves.

0:28:47 > 0:28:49Just like we do.

0:28:49 > 0:28:52So imitation learning is the process of having a robot learn from

0:28:52 > 0:28:54watching something done for it.

0:28:54 > 0:28:57Then in the future, faced with a different situation,

0:28:57 > 0:29:00as to tie a knot, we're going to see the robot

0:29:00 > 0:29:03understand what to do in the new situation.

0:29:03 > 0:29:04For this challenge,

0:29:04 > 0:29:08the rope is relayed in roughly the same place and Brett is asked to tie

0:29:08 > 0:29:10the knot again.

0:29:10 > 0:29:12Right, let's see what Brett can do now.

0:29:14 > 0:29:17Brett can't simply repeat the exact motions he's been shown because the

0:29:17 > 0:29:19rope isn't in exactly the same place.

0:29:21 > 0:29:22He'll have to adapt the principles

0:29:22 > 0:29:26of the challenge to the reality of the rope's new position.

0:29:27 > 0:29:30It's not just repeating the motions we gave.

0:29:30 > 0:29:31It's looking at how the new

0:29:31 > 0:29:34situation relates to the old situation and

0:29:34 > 0:29:36then morphing the motions,

0:29:36 > 0:29:38make them the right motions for the new situation.

0:29:38 > 0:29:40Yeah, there, Brett did it.

0:29:40 > 0:29:42Beautiful knot.

0:29:42 > 0:29:46Beautiful indeed, but Brett's skills don't end at imitation.

0:29:48 > 0:29:52He can work out how to perform a task from scratch, like a toddler.

0:29:53 > 0:29:58A rather unattractive toddler with metal arms and an electronic brain.

0:29:58 > 0:30:02But if he can learn, there's no reason why one day Brett,

0:30:02 > 0:30:04or a robot like him, couldn't learn to drive.

0:30:06 > 0:30:09But we're getting ahead of ourselves.

0:30:09 > 0:30:11We'll specify the objective rather

0:30:11 > 0:30:13than the strategy to achieve something.

0:30:13 > 0:30:14So you don't have to demonstrate

0:30:14 > 0:30:17everything multiple times for the robot to understand what to do.

0:30:17 > 0:30:19You just give the robot an objective.

0:30:19 > 0:30:22Then the robot just can go at it,

0:30:22 > 0:30:23pretty much like a kid playing,

0:30:23 > 0:30:26trying, trying, trying and, over time,

0:30:26 > 0:30:27getting better and better.

0:30:29 > 0:30:31This challenge requires Brett to get

0:30:31 > 0:30:35the red cube into the box through one of the holes.

0:30:35 > 0:30:36But he's not shown how.

0:30:38 > 0:30:40He has no model of how his own arm works,

0:30:40 > 0:30:43so initially the best he can do is random motion.

0:30:43 > 0:30:44So right now...

0:30:46 > 0:30:49As it refines that model and gets more data from its own attempts it

0:30:49 > 0:30:50gets better and better at finding a

0:30:50 > 0:30:53solution to get the block into the matching opening.

0:30:53 > 0:30:55And actually, very quickly.

0:30:55 > 0:30:56This took maybe one minute of learning

0:30:56 > 0:30:59and it invented how to do that from scratch.

0:30:59 > 0:31:01With imitation learning,

0:31:01 > 0:31:03Brett simply adapts an action that

0:31:03 > 0:31:06he's been shown, but machine learning,

0:31:06 > 0:31:10artificial intelligence, allows him to solve a problem for himself.

0:31:12 > 0:31:16The implications for his descendants and ours are far-reaching.

0:31:18 > 0:31:21Maybe you want a robot to jump higher than a human has ever jumped.

0:31:21 > 0:31:24Or you want to do something more precise than a human has ever done.

0:31:24 > 0:31:28You can give it the objective and then it'll, on its own, try.

0:31:28 > 0:31:30Not succeed initially, fail most of the time,

0:31:30 > 0:31:34but over time get better and better and maybe exceed human performance.

0:31:36 > 0:31:38When it comes to driverless cars,

0:31:38 > 0:31:41exceeding human performance is essential.

0:31:43 > 0:31:46After all, not many of us would want to be driven around by Brett.

0:31:52 > 0:31:54Artificial intelligence is the

0:31:54 > 0:31:57latest technological new kid on the block

0:31:57 > 0:31:58and areas that seemingly have

0:31:58 > 0:32:02nothing to do with driving are busy making

0:32:02 > 0:32:04friends with it, unwittingly

0:32:04 > 0:32:07preparing AI for its upcoming automotive challenge.

0:32:09 > 0:32:13Ruby, for example, has memorised more than 100 billion web pages.

0:32:13 > 0:32:15No human can do this. In fact,

0:32:15 > 0:32:16it can find the right web page as

0:32:16 > 0:32:19you're still typing your search term.

0:32:19 > 0:32:22That is completely impossible for the human brain.

0:32:22 > 0:32:23I mean, before machine learning we

0:32:23 > 0:32:26used to program computers line by line.

0:32:26 > 0:32:30But now, we can teach computers and let them learn on their own,

0:32:30 > 0:32:31the same way people learn.

0:32:33 > 0:32:36At the social media giant Facebook,

0:32:36 > 0:32:39artificial intelligence underpins the whole operation.

0:32:41 > 0:32:46Facebook today could not exist without AI. It's as simple as that.

0:32:46 > 0:32:48Over a billion people use Facebook

0:32:48 > 0:32:50every day and they will load their news

0:32:50 > 0:32:52feed a few dozen times every single day.

0:32:52 > 0:32:54And, you know, if you imagine how

0:32:54 > 0:32:56many people it would take if you were to

0:32:56 > 0:32:59line out all the pieces of content that are available to you every

0:32:59 > 0:33:00day, for how to sort out how

0:33:00 > 0:33:03relevant is this story going to be to

0:33:03 > 0:33:05this person, you multiply that by a billion people,

0:33:05 > 0:33:07you see that for a human,

0:33:07 > 0:33:10this would be a task that is absolutely impossible to do.

0:33:11 > 0:33:14Software engineers are trying to achieve artificial intelligence by

0:33:14 > 0:33:18programming computers to process information like the human brain.

0:33:20 > 0:33:23They have come up with artificial neural networks,

0:33:23 > 0:33:26systems that allow what researchers call deep learning.

0:33:28 > 0:33:30My name is Yann Lecun, I run and I

0:33:30 > 0:33:32build a big research group at Facebook,

0:33:32 > 0:33:35discovering new things that nobody has thought about before.

0:33:35 > 0:33:38So when you talk to your phone and it recognises your voice,

0:33:38 > 0:33:39it uses deep learning. When you

0:33:39 > 0:33:41upload your photos on Facebook and Google

0:33:41 > 0:33:43and you can index them and search them,

0:33:43 > 0:33:46they've been recognised by deep learning.

0:33:48 > 0:33:51Deep learning mimics the way we learn ourselves.

0:33:53 > 0:33:55In our brains, this is done by

0:33:55 > 0:33:58strengthening and weakening neural connections.

0:33:58 > 0:34:00Yann has designed an artificial

0:34:00 > 0:34:03system that does a similar thing in the virtual world.

0:34:08 > 0:34:10When you get an image in your eyes, on your retina,

0:34:10 > 0:34:12it goes to the back of the brain and then it's kind of processed by

0:34:12 > 0:34:14multiple layers of neurons.

0:34:14 > 0:34:17And that process takes about 100 milliseconds and it goes through

0:34:17 > 0:34:19only a few layers of neurons, maybe between ten and 20.

0:34:19 > 0:34:21So these neural nets that we're building,

0:34:21 > 0:34:24they're called convolutional networks.

0:34:24 > 0:34:25It's a little bit my invention and

0:34:25 > 0:34:28they're organised in a very similar way.

0:34:29 > 0:34:32The reason that people are excited about Yann's convolutional,

0:34:32 > 0:34:34artificial neural networks is their

0:34:34 > 0:34:37astonishing abilities in object recognition.

0:34:39 > 0:34:40They can be taught, for example,

0:34:40 > 0:34:43to look at the digital data from a photograph

0:34:43 > 0:34:46and let you know whether or not the picture contains a dog.

0:34:49 > 0:34:52This is done by showing the system lots of pictures of dogs,

0:34:52 > 0:34:55while telling the artificial neural network that these are dogs.

0:34:57 > 0:35:00By looking for common features in the pictures,

0:35:00 > 0:35:04the system will develop its own definition of what constitutes dog,

0:35:04 > 0:35:05so that eventually,

0:35:05 > 0:35:07it will be able to spot a dog or the

0:35:07 > 0:35:09fact that a picture doesn't contain

0:35:09 > 0:35:11a dog, all by itself.

0:35:12 > 0:35:17Of course, it's not just dogs. It could be anything - cars, people,

0:35:17 > 0:35:19penguins, oncoming traffic.

0:35:21 > 0:35:24It's actually better at doing this

0:35:24 > 0:35:27than a human engineer would be at designing a system.

0:35:27 > 0:35:28That's the surprising thing, it's

0:35:28 > 0:35:31very humbling for an engineer to think that's the case, but it is.

0:35:31 > 0:35:32And they work really well for image

0:35:32 > 0:35:35recognition, for video interpretation.

0:35:35 > 0:35:37So all the self-driving cars that

0:35:37 > 0:35:39you see around that use a camera input,

0:35:39 > 0:35:40they all use convolutional nets.

0:35:40 > 0:35:42It's these systems' ability to

0:35:42 > 0:35:44differentiate at a better than human level

0:35:44 > 0:35:46which is at the heart of artificial intelligence.

0:35:48 > 0:35:50It's what powers Brett,

0:35:50 > 0:35:53the knot-tying robot, and it's going to be critical to the difference

0:35:53 > 0:35:56between level three driverless cars that almost work

0:35:56 > 0:35:59and level four driverless cars that really do.

0:36:01 > 0:36:02But it's not the only thing.

0:36:03 > 0:36:05The other thing is this.

0:36:06 > 0:36:10Video games are massively demanding on computer power,

0:36:10 > 0:36:12so much so that a whole new way of

0:36:12 > 0:36:15processing is needed to handle all the data they use.

0:36:17 > 0:36:22What evolved was this - the graphics processing unit or GPU.

0:36:24 > 0:36:27And because of their data handling capabilities,

0:36:27 > 0:36:29GPUs have been gleefully adopted by

0:36:29 > 0:36:32the would-be makers of driverless cars.

0:36:34 > 0:36:36My name's Danny Shapiro.

0:36:36 > 0:36:38I'm the Senior Director of Automotive at Nvidia

0:36:38 > 0:36:40and we're building systems

0:36:40 > 0:36:42to enable cars to drive themselves.

0:36:44 > 0:36:47This is the brain of a self-driving car.

0:36:47 > 0:36:49We plug in cameras, radar,

0:36:49 > 0:36:54lidar. All the sensors of the car feed into this device.

0:36:55 > 0:36:57A CPU, which is the central processing unit,

0:36:57 > 0:37:02you've probably heard, has dual core or quad core,

0:37:02 > 0:37:07meaning there's two lanes or four lanes where information flows.

0:37:07 > 0:37:10The GPU can have thousands of cores or lanes.

0:37:10 > 0:37:13Imagine a highway with 1,000 lanes,

0:37:13 > 0:37:15how much traffic could you push through that processor?

0:37:20 > 0:37:23Just like the brains of human drivers,

0:37:23 > 0:37:27the systems that control driverless cars will be

0:37:27 > 0:37:29voracious consumers of data.

0:37:29 > 0:37:35They will be fed with a constant stream of digits from lidar, radar,

0:37:35 > 0:37:37infrared sensors and multiple video cameras,

0:37:37 > 0:37:40all of which will need to be seamlessly interpreted,

0:37:40 > 0:37:44coordinated and fed back in the form of different data to the car's

0:37:44 > 0:37:47driving controls in real time.

0:37:47 > 0:37:51The driverless car has to use all this to sense and interpret the real

0:37:51 > 0:37:54world with 100% accuracy.

0:37:57 > 0:38:00A GPU is able to process and reconstruct, essentially,

0:38:00 > 0:38:03a three-dimensional model of everything going on around the car.

0:38:03 > 0:38:06All that data, then, is analysed.

0:38:06 > 0:38:08It doesn't just sense there's an object,

0:38:08 > 0:38:10but we know exactly what that object is.

0:38:10 > 0:38:13It could be a pedestrian on a cellphone,

0:38:13 > 0:38:16it could be a motorcycle, it could be an ambulance.

0:38:16 > 0:38:18British start-up FIVE AI are training

0:38:18 > 0:38:20their car to recognise things too.

0:38:21 > 0:38:23Formula 1 engineer Teena Gade

0:38:23 > 0:38:26is being shown the world through the eyes of the driverless car.

0:38:28 > 0:38:31So talk me through what we've got on the screen here.

0:38:31 > 0:38:33The computer vision is making the machine see.

0:38:33 > 0:38:38So here we have the live stream from the camera coming in and these are

0:38:38 > 0:38:41representations that have been processed from that.

0:38:41 > 0:38:44So in the first one, what we're seeing here is in real time,

0:38:44 > 0:38:46the actual detection of buses, cars,

0:38:46 > 0:38:50pedestrians that have been inferred by our algorithm.

0:38:50 > 0:38:53The next one is what we'd call a segmentation,

0:38:53 > 0:38:56which is a breaking-up of the image into something like here's where the road

0:38:56 > 0:39:00is, here's a wall, here's a building,

0:39:00 > 0:39:04so that the car has a very coarse awareness of what's around it.

0:39:04 > 0:39:06And the same sorts of machine learning techniques,

0:39:06 > 0:39:09neural networks is what will also

0:39:09 > 0:39:11help you solve looking at intentions of

0:39:11 > 0:39:14other road users. You can imagine, just as a human goes out,

0:39:14 > 0:39:17they learn how other road users use by observation.

0:39:17 > 0:39:19So you'd feed it millions and

0:39:19 > 0:39:23millions of days of video in all different situations.

0:39:23 > 0:39:27And the machine itself would learn how to understand their

0:39:27 > 0:39:30movements, maybe picking up cues that to you and I,

0:39:30 > 0:39:31we would never have even thought of,

0:39:31 > 0:39:36because it would have so much data at its disposal.

0:39:36 > 0:39:39But there is a fascinating short cut to this,

0:39:39 > 0:39:41one that Teena exploits in her work.

0:39:41 > 0:39:43You might want to turn round.

0:39:43 > 0:39:45I don't know what's up here.

0:39:45 > 0:39:48And one that FIVE AI are making full use of.

0:39:48 > 0:39:50This, actually, looks quite familiar to me.

0:39:50 > 0:39:52This is a simulated environment,

0:39:52 > 0:39:54and that's how we test our cars on the track.

0:39:54 > 0:39:58Yes. So one of the exciting things about the development in gaming

0:39:58 > 0:40:01engines and simulations of reality is they're getting so good...

0:40:01 > 0:40:05I mean, if you look at, say, Grand Theft Auto and games like this,

0:40:05 > 0:40:07it's now become possible to

0:40:07 > 0:40:10actually do testing in these virtual worlds

0:40:10 > 0:40:12which is almost as good as testing in reality.

0:40:12 > 0:40:14And you could get algorithms up to a

0:40:14 > 0:40:17sufficient case on very rare test cases

0:40:17 > 0:40:19that you just wouldn't have access to.

0:40:19 > 0:40:21You know, creating accidents, things like this,

0:40:21 > 0:40:25which you wouldn't want to do in the real world, and actually, you know,

0:40:25 > 0:40:28it takes a lot of money to run a car out in the real world for, like,

0:40:28 > 0:40:3090 million miles.

0:40:30 > 0:40:33If you have enough computers, you can do it in the virtual world

0:40:33 > 0:40:35in, you know, hours.

0:40:36 > 0:40:38It's the modern blurring of the real

0:40:38 > 0:40:41with the digitally virtual that means

0:40:41 > 0:40:45that now, right now, in the early 21st century,

0:40:45 > 0:40:48it might be that the driverless car's time has come.

0:40:51 > 0:40:54Most of the ideas about AI have been around for quite a while.

0:40:54 > 0:40:56What we now have is the technology.

0:40:56 > 0:40:57So we have the computational power

0:40:57 > 0:40:59to run them as well as the ability to

0:40:59 > 0:41:01gather the data in order to learn

0:41:01 > 0:41:03what we need to learn and to train these systems.

0:41:03 > 0:41:06So we're at this point now where driverless cars and driving

0:41:06 > 0:41:08simulators can do both of these things,

0:41:08 > 0:41:11because the technology enables us to at the moment.

0:41:11 > 0:41:14And that's certainly what the car industry believes.

0:41:14 > 0:41:16Welcome to the world of level four.

0:41:19 > 0:41:21Level four.

0:41:21 > 0:41:23According to the plan,

0:41:23 > 0:41:25a level four car will perform all

0:41:25 > 0:41:28driving functions and monitor roadway

0:41:28 > 0:41:30conditions for an entire trip within

0:41:30 > 0:41:33the operational designed domain of the vehicle.

0:41:35 > 0:41:37So this is it.

0:41:37 > 0:41:40This is, to all intents and purposes, the future...

0:41:41 > 0:41:45..driverless cars that will ferry you around with no need for you to worry

0:41:45 > 0:41:49yourself with troublesome bits of kit like steering wheels, or brakes,

0:41:49 > 0:41:53or gears, or anything, really.

0:41:53 > 0:41:56So just imagine, you go to your smartphone, you say,

0:41:56 > 0:42:00"I need a ride to the pizza place," and you punch in the app,

0:42:00 > 0:42:03and the car pops up in front of your house, empty.

0:42:03 > 0:42:06You hop inside, it drops you at the restaurant, you have dinner,

0:42:06 > 0:42:10you drink a lot, you are reasonably drunk now, OK?

0:42:10 > 0:42:12And you will go home and do the same thing again,

0:42:12 > 0:42:14and the car safely brings you home.

0:42:14 > 0:42:16That's going to happen in the next five years.

0:42:16 > 0:42:19So the future's bright for pizza-eating alcoholics,

0:42:19 > 0:42:20so long as they live in the

0:42:20 > 0:42:22operational designed domain of their vehicle.

0:42:24 > 0:42:29But within these zones, level four does offer complete autonomy,

0:42:29 > 0:42:32with no driving controls available to passengers at all.

0:42:34 > 0:42:37However unlikely this vision of the future may seem,

0:42:37 > 0:42:39this is exactly what some

0:42:39 > 0:42:43manufacturers are promising us is just around the corner.

0:42:43 > 0:42:49We're announcing Ford's intent to have a high-volume, SAE, level four,

0:42:49 > 0:42:52fully autonomous vehicle in

0:42:52 > 0:42:57commercial operation in 2021 in a ride-hailing

0:42:57 > 0:42:59or ride-sharing service.

0:43:00 > 0:43:02It's a bold claim,

0:43:02 > 0:43:06and one that will test the navigational systems to their limit.

0:43:06 > 0:43:08But even if these machines come good,

0:43:08 > 0:43:10there's another subtle aspect to

0:43:10 > 0:43:13driving that is sometimes overlooked.

0:43:13 > 0:43:15Probably because it's such a human issue.

0:43:22 > 0:43:26This small autonomous machine is called Jack Rabot.

0:43:27 > 0:43:30By going about his smartly dressed business,

0:43:30 > 0:43:32he's finding out how people behave around each other.

0:43:35 > 0:43:40Jack Rabot learns from the behaviour of other people how to move around.

0:43:40 > 0:43:44The more he looks at people, the better he will be in his navigation.

0:43:45 > 0:43:49It turns out that we humans, when we navigate in crowded scene,

0:43:49 > 0:43:53we read each other's behaviour, body language to avoid each other.

0:43:53 > 0:43:58We respect personal space, yield right of way, and the culture,

0:43:58 > 0:44:02the way people decide to move around is different,

0:44:02 > 0:44:06because we all have different social behaviour,

0:44:06 > 0:44:09and this behaviour can only be learned from the observation,

0:44:09 > 0:44:12from the data. We cannot define them.

0:44:12 > 0:44:14We cannot

0:44:14 > 0:44:15write the rules that should be

0:44:15 > 0:44:19applied in the UK and in Japan on the same time.

0:44:19 > 0:44:21That's because, however adept Jack

0:44:21 > 0:44:24might be at schmoozing with Californian students,

0:44:24 > 0:44:28abandon him in, say, Dagenham, and he might struggle.

0:44:28 > 0:44:30It's the same with driving.

0:44:30 > 0:44:33The reason is that

0:44:33 > 0:44:35the way Americans drive is not the

0:44:35 > 0:44:39same as French and British people drive.

0:44:42 > 0:44:44The only limiting factor of a level

0:44:44 > 0:44:48four car is that it will only work in predefined areas.

0:44:49 > 0:44:53A level five vehicle will be able to work anywhere,

0:44:53 > 0:44:55whatever the driving conditions and

0:44:55 > 0:44:58whatever the cultural driving conventions.

0:44:58 > 0:45:00It might be, then,

0:45:00 > 0:45:04that Jack is helping to deliver the ultimate driverless nirvana.

0:45:06 > 0:45:08Level five.

0:45:08 > 0:45:11We have now arrived at full autonomy.

0:45:11 > 0:45:13Now, according to the masterplan,

0:45:13 > 0:45:17the car will have capabilities at least equal to a human driver in

0:45:17 > 0:45:19every possible driving scenario.

0:45:22 > 0:45:25What this actually means is what it says.

0:45:25 > 0:45:27Pop in a postcode or, if you're more rugged,

0:45:27 > 0:45:31GPS coordinates, and off you go, anywhere,

0:45:31 > 0:45:33and do anything you like on the way.

0:45:35 > 0:45:37At FIVE AI,

0:45:37 > 0:45:41Teena is hoping to witness the reasonably priced car's first steps

0:45:41 > 0:45:44on the path to this full level five autonomy.

0:45:45 > 0:45:47OK, so this is basically the first day in the real world.

0:45:47 > 0:45:49It is the first day in the real world, indeed.

0:45:49 > 0:45:51Yeah. So we're at a test track here,

0:45:51 > 0:45:54so we're not going to sort of hopefully damage anything.

0:45:54 > 0:45:56Excellent. Shall we go, then?

0:45:56 > 0:45:58Let's go.

0:45:58 > 0:46:01So what I'm going to do here... We're going to just hit return.

0:46:01 > 0:46:02OK.

0:46:03 > 0:46:05We're now off. I've got overall

0:46:05 > 0:46:07control of the car with this dead man's

0:46:07 > 0:46:10handle here, so if there's anything that goes disastrously wrong,

0:46:10 > 0:46:14we can always stop. But as you can see, we're not going very fast,

0:46:14 > 0:46:16actually, we're going about four or five miles an hour.

0:46:16 > 0:46:18So it's... There's not a huge danger

0:46:18 > 0:46:20of anything particularly going wrong.

0:46:22 > 0:46:23Within one or two days, we'll be

0:46:23 > 0:46:26going round much more complex tracks,

0:46:26 > 0:46:27and within a few weeks,

0:46:27 > 0:46:32we'll be able to deal with simple kind of junctions and certainly able

0:46:32 > 0:46:33to deal with obstacles.

0:46:34 > 0:46:38Actually turned out to be OK. Looked like it was avoiding those cones.

0:46:38 > 0:46:40It did look like it was avoiding those cones.

0:46:40 > 0:46:42Although we weren't fully sure.

0:46:42 > 0:46:43We weren't fully sure.

0:46:45 > 0:46:46It's early days,

0:46:46 > 0:46:50but this particular electric car has got a fair way to go if it's to

0:46:50 > 0:46:51achieve level five autonomy.

0:46:55 > 0:46:59But if it does, it could revolutionise our driving world.

0:47:02 > 0:47:04I'm excited by the prospect of driverless cars.

0:47:04 > 0:47:07You'll just sit in the cab and it will take you where you want to go.

0:47:07 > 0:47:10And to me, that's good. I can do other things.

0:47:10 > 0:47:12Sure, it'd be a big deal to be able to get into your car and curl up in

0:47:12 > 0:47:16the back seat and have a nap while you go from A to B, but

0:47:16 > 0:47:19if you think about how cities and whole countries are built around the

0:47:19 > 0:47:21road network,

0:47:21 > 0:47:24the changes that could happen to that are enormous.

0:47:24 > 0:47:27And that's really the challenge -

0:47:27 > 0:47:30what does the infrastructure need to look like for these vehicles?

0:47:36 > 0:47:39One major challenge is that, for the past century or so,

0:47:39 > 0:47:43we've been building our world around a totally different kind of car.

0:47:43 > 0:47:46And before you even think about autonomy,

0:47:46 > 0:47:48you need to gear up for electric.

0:47:50 > 0:47:52My name's Gareth Dunsmore.

0:47:52 > 0:47:54I run electric vehicles for Nissan in Europe.

0:47:54 > 0:47:56I perhaps enjoy driving

0:47:56 > 0:47:58a Leaf more than any vehicle I've driven,

0:47:58 > 0:48:01an electric vehicle more than any other vehicle I've driven.

0:48:01 > 0:48:03- Really?- Yeah.- You just come here, you have to say that, don't you?

0:48:03 > 0:48:06No. It's different, it's a different driving experience.

0:48:06 > 0:48:09If you've not done it you can't explain it, but it's...

0:48:09 > 0:48:10It's instant acceleration,

0:48:10 > 0:48:14and that... You don't just... You don't get that even in a GTR.

0:48:14 > 0:48:15Apart from the obvious thrill of

0:48:15 > 0:48:18driving Nissan's entry-level electric

0:48:18 > 0:48:21option, Gareth has another reason to be evangelical

0:48:21 > 0:48:23about alternative power.

0:48:23 > 0:48:27Vehicles are far easier to automate if they are also electric.

0:48:27 > 0:48:30On top of that, though, I think there's a broader point.

0:48:30 > 0:48:33Looking at cities and looking at what we're trying to achieve,

0:48:33 > 0:48:37it's about moving to zero emissions and zero fatalities.

0:48:37 > 0:48:40So combining the two technologies together makes

0:48:40 > 0:48:43more sense from a customer perspective,

0:48:43 > 0:48:45to be able to bring forward both at the same time.

0:48:47 > 0:48:49In planning for this electric utopia,

0:48:49 > 0:48:53Gareth teamed up with British architects Foster + Partners.

0:48:53 > 0:48:55Together, they came up with a

0:48:55 > 0:48:57version of the future that imagines how

0:48:57 > 0:49:01electric autonomous cars might do more than just drive themselves.

0:49:03 > 0:49:05I think the challenge

0:49:05 > 0:49:08is integrating it into, you know, existing city fabric.

0:49:08 > 0:49:11And that's where the conversations with Nissan about taking one part of

0:49:11 > 0:49:14it and the charging strategy and autonomously doing that...

0:49:14 > 0:49:17They're all going to have their own nuances which need to be solved.

0:49:17 > 0:49:19I think it's shifted our perspective slightly.

0:49:19 > 0:49:23We kind of, you know...probably would have tackled the problem in a

0:49:23 > 0:49:25very, very different way. But you realise

0:49:25 > 0:49:27the issues facing urban development around the world,

0:49:27 > 0:49:31you need a very integrated approach to transport, in particular,

0:49:31 > 0:49:33how you move from public to private transport,

0:49:33 > 0:49:37how you cover big distances, what's the economics of all of that,

0:49:37 > 0:49:39how do you make it work?

0:49:39 > 0:49:43With Nissan's wireless charging and universal connectivity,

0:49:43 > 0:49:47our vehicles could autonomously charge themselves

0:49:47 > 0:49:49and then re-park so another vehicle

0:49:49 > 0:49:52on the street could use the same bay,

0:49:52 > 0:49:54all while you sleep.

0:49:54 > 0:49:57I think the thing that we're quite excited about is

0:49:57 > 0:50:00there's less reliance on having your car always parked outside your

0:50:00 > 0:50:04house. Once we release them, your car can come from round the corner.

0:50:04 > 0:50:07We can start opening up some of these sort of residential areas,

0:50:07 > 0:50:09and they'll feel much, you know...

0:50:09 > 0:50:12It almost goes back to, you know, a hundred years ago, streets

0:50:12 > 0:50:14become actually a really nice place to be outside,

0:50:14 > 0:50:17not in your little sort of hidden gardens.

0:50:17 > 0:50:18And we become more sociable again and

0:50:18 > 0:50:20use these spaces that are more shared.

0:50:22 > 0:50:25Self-driving cars are going to have a huge impact.

0:50:25 > 0:50:28If all cars are self-driving, we can get rid of the streetlights,

0:50:28 > 0:50:31cos they won't need them, they'll be able to navigate through complex

0:50:31 > 0:50:33intersections with no collisions.

0:50:34 > 0:50:37Right now, we have in the United States 100 million cars.

0:50:37 > 0:50:41They are parked 97% of the time, only driven 3%.

0:50:41 > 0:50:44So in the future, we're going to have less traffic,

0:50:44 > 0:50:48and once we have robotic, self-driving car taxi services,

0:50:48 > 0:50:50we don't need parking spaces any more.

0:50:50 > 0:50:52It means that the cities will look nicer.

0:50:52 > 0:50:56If cars are not crashing, we're not going to need as many doctors,

0:50:56 > 0:50:59and we can totally reimagine the car.

0:50:59 > 0:51:03We don't need the heavy steel, rigid bodies of these cars to protect the

0:51:03 > 0:51:06inhabitants. They could be made of more environmental friendly,

0:51:06 > 0:51:08more lightweight materials.

0:51:08 > 0:51:12So great societal benefits with self-driving cars.

0:51:14 > 0:51:15Well, that's settled, then.

0:51:15 > 0:51:17Driverless cars will usher in a

0:51:17 > 0:51:20world where road traffic accidents will be

0:51:20 > 0:51:23a thing of the past, where the lion will lie down with the lamb,

0:51:23 > 0:51:26swords will be beaten into ploughshares,

0:51:26 > 0:51:28and people will be nicer, kinder,

0:51:28 > 0:51:30happier and richer.

0:51:31 > 0:51:33Except, of course, they won't.

0:51:34 > 0:51:36I think a wholly autonomous,

0:51:36 > 0:51:40driverless future could have more social impacts than we imagine.

0:51:40 > 0:51:41I mean, first off, you've got to

0:51:41 > 0:51:44think about all the people who drive to make a living.

0:51:44 > 0:51:46And so you've got to think about how automation will affect

0:51:46 > 0:51:49their livelihood.

0:51:49 > 0:51:52There are about half a million taxi drivers, delivery drivers

0:51:52 > 0:51:54and bus drivers in the UK,

0:51:54 > 0:51:57who might well need to look for alternative employment.

0:51:58 > 0:52:00But as if that wasn't bad enough,

0:52:00 > 0:52:02driverless cars might even make our

0:52:02 > 0:52:04cities' congestion problems even worse.

0:52:05 > 0:52:09I can see very quickly a time where people won't actually stop their car

0:52:09 > 0:52:11driving when they want to go to the shops.

0:52:11 > 0:52:13They'll go to the shops, they'll get out,

0:52:13 > 0:52:15and then they'll tell their car to drive around the block and wait till

0:52:15 > 0:52:17they're ready.

0:52:17 > 0:52:20And the problems don't stop there.

0:52:20 > 0:52:23There are practical and ethical issues yet to be resolved.

0:52:24 > 0:52:27It'll be difficult for autonomous cars to coexist with people driving

0:52:27 > 0:52:31cars, which will obviously be the case for many years to come.

0:52:31 > 0:52:32The cars with drivers are

0:52:32 > 0:52:35potentially going to take advantage of the

0:52:35 > 0:52:38driverless cars, which they know have to stop and give way when their

0:52:38 > 0:52:40sensors detect that something's wrong.

0:52:43 > 0:52:46What happens when it eventually encounters a no-win scenario,

0:52:46 > 0:52:50when it actually has to have an accident, where it cannot avoid it?

0:52:50 > 0:52:56If you get to the point where the car has to choose between

0:52:56 > 0:53:01an old man in a vehicle or a young lady and an infant in the other car,

0:53:01 > 0:53:03and it knows it's going to hit one of them,

0:53:03 > 0:53:06these cars are going to have to be taught to make these moral,

0:53:06 > 0:53:10ethical judgments that you make in a snap decision.

0:53:10 > 0:53:13Some car manufacturers have already said that they're going to teach the

0:53:13 > 0:53:17car to favour the driver above all else.

0:53:17 > 0:53:19Now, what if you don't want that?

0:53:19 > 0:53:22You might actually set those parameters in the vehicle yourself,

0:53:22 > 0:53:24the level of morality for the car.

0:53:27 > 0:53:30Where some driver-assist technologies have been introduced,

0:53:30 > 0:53:33there have already been problems.

0:53:33 > 0:53:35A Tesla crashed, killing its driver,

0:53:35 > 0:53:37and Uber suspended their fleet of

0:53:37 > 0:53:42semi-autonomous test cars following a non-fatal collision.

0:53:42 > 0:53:45But despite this, the technology moves forward.

0:53:53 > 0:53:56It's three weeks since Teena was at FIVE AI, and today,

0:53:56 > 0:53:58she's going back to meet them for the last time,

0:53:58 > 0:54:00to see how their car's getting on

0:54:00 > 0:54:02with its assault on level five autonomy.

0:54:06 > 0:54:09What we're now going to do is just engage our robot.

0:54:09 > 0:54:12Press that button there. Robot is now in control.

0:54:12 > 0:54:14And off we go.

0:54:14 > 0:54:17OK, so the car's now setting off.

0:54:17 > 0:54:19See it's quickly got up to...

0:54:19 > 0:54:21Doing about 12 miles an hour on this access road here.

0:54:21 > 0:54:23So right now, we're driving entirely autonomous.

0:54:23 > 0:54:25We're driving entirely autonomously.

0:54:25 > 0:54:28Actually, this robot's getting a bit rattly.

0:54:28 > 0:54:30- It is.- A few of the things we've

0:54:30 > 0:54:32really improved over the last few weeks

0:54:32 > 0:54:35is we've managed to sort out a lot of the visual odometry,

0:54:35 > 0:54:36so the car can work out really

0:54:36 > 0:54:39accurately where it is in three-dimensional

0:54:39 > 0:54:43space. Just from the cameras and the pixels on the camera and their

0:54:43 > 0:54:45- motion...- Yeah.- ..it can actually work out...

0:54:45 > 0:54:48It can in fact go round really quite long tracks,

0:54:48 > 0:54:51and within a few centimetres can detect exactly where it is in

0:54:51 > 0:54:54three-dimensional space without the use of a map.

0:54:54 > 0:54:56- Yeah.- So that's pretty cool.

0:54:56 > 0:54:59Although, not quite cool enough.

0:54:59 > 0:55:01An old-fashioned level zero

0:55:01 > 0:55:04interception is required to save the car from itself.

0:55:04 > 0:55:06Whoa, it's too close.

0:55:06 > 0:55:10Let's just make sure this thing is not going to hit that barrier there.

0:55:10 > 0:55:13There's some latencies in the system that we're now going to iron out.

0:55:13 > 0:55:15And when you say latency, what's actually happening

0:55:15 > 0:55:17is the car's not responding as fast as you'd like.

0:55:17 > 0:55:19That's right. Yeah, it takes...

0:55:19 > 0:55:21It takes a few hundreds of

0:55:21 > 0:55:23milliseconds to go all the way through from

0:55:23 > 0:55:25seeing data in the cameras to

0:55:25 > 0:55:29actually a decision that actually controls the wheel.

0:55:29 > 0:55:31And that gap between the two

0:55:31 > 0:55:34means that by the time we actually apply control,

0:55:34 > 0:55:37it's to a situation that was true maybe half a second ago.

0:55:37 > 0:55:39- Right.- And, you know,

0:55:39 > 0:55:40we need to take all that into

0:55:40 > 0:55:42account in the way we design the car.

0:55:42 > 0:55:44Also the way we speed up some of the algorithms, as well.

0:55:44 > 0:55:47- Yeah.- Stop it there at that point, actually.

0:55:47 > 0:55:49We're about to hit the roundabout. So just go round this roundabout.

0:55:49 > 0:55:53Basically, it is still in a slightly drunken manner, but...

0:55:53 > 0:55:56Yeah, it is... It is suffering from a serious case of PIO.

0:55:56 > 0:55:58Pilot-induced oscillation.

0:55:58 > 0:56:00Responding to the signal too late.

0:56:00 > 0:56:02Very well documented in lots of aeronautics texts.

0:56:02 > 0:56:04Well, there you go. You know,

0:56:04 > 0:56:06this is why we should have cross-disciplinary...

0:56:06 > 0:56:08Yeah, you need some dynamicists on board.

0:56:08 > 0:56:12We do need some dynamicists, I tell you. Yeah.

0:56:12 > 0:56:15Despite the obvious challenges, Stan remains optimistic.

0:56:18 > 0:56:22In 2019, we're going to be driving autonomously in urban scenes,

0:56:22 > 0:56:25and I reckon there's going to be commercial, you know,

0:56:25 > 0:56:27within one-and-a-half to two years after that.

0:56:28 > 0:56:32FIVE AI have chosen the hardest problem to solve,

0:56:32 > 0:56:34that of complete autonomy,

0:56:34 > 0:56:37so their journey in reaching that goal could well be a long one.

0:56:44 > 0:56:46The prospect of replacing our cars

0:56:46 > 0:56:49with driverless ones tends to split opinion.

0:56:51 > 0:56:52You know, we've done sort of surveys

0:56:52 > 0:56:55and focus groups and things like that

0:56:55 > 0:56:58with people, and it's honestly 50-50.

0:56:58 > 0:57:00Some people hate the idea of

0:57:00 > 0:57:02relinquishing the pleasure of driving.

0:57:02 > 0:57:05I love driving. I'm an automotive journalist,

0:57:05 > 0:57:07so I have to love driving or I wouldn't do my job.

0:57:07 > 0:57:09Others relish the prospect of

0:57:09 > 0:57:12freeing up wasted time behind the wheel.

0:57:12 > 0:57:15If it can make my commute into work that much easier,

0:57:15 > 0:57:18I won't have to stress about it in the morning, that'd be grand.

0:57:18 > 0:57:21If I could just sit back and read a book, listen to some music,

0:57:21 > 0:57:24catch up on some sleep, that would be great.

0:57:24 > 0:57:26For some, it's the ultimate freedom.

0:57:26 > 0:57:30For others, it's an opportunity for less freedom, for a longer work day.

0:57:32 > 0:57:34All those times when you were travelling,

0:57:34 > 0:57:38and it was a break from the phone calls and it was a break from the

0:57:38 > 0:57:43e-mail, I think those times would be freed up for us to actually carry on

0:57:43 > 0:57:46working and the daily grind.

0:57:47 > 0:57:52But despite the reservations, the die seems to have been cast.

0:57:52 > 0:57:55The dawn of the driverless car is here.

0:57:55 > 0:57:59The challenge for the technologists is to make sure that the transition

0:57:59 > 0:58:04into reality is as beguilingly smooth as the PR that surrounds it.

0:58:04 > 0:58:08I can't foresee the future, but I can build it.

0:58:08 > 0:58:11The reason that I believe it's good to be a technology optimist is

0:58:11 > 0:58:13throughout the entire history of the human race,

0:58:13 > 0:58:15technology has empowered us.

0:58:15 > 0:58:19From the very early days, the Bronze Age, the Stone Age,

0:58:19 > 0:58:23to the day of the smartphones and modern medicine, it has freed us,

0:58:23 > 0:58:25it has levelled the playing field for everybody,

0:58:25 > 0:58:27and has empowered us as a human race.

0:58:27 > 0:58:28Why stop that?