WEBVTT - How do self-driving cars work?

0:00:00.520 --> 0:00:03.840
<v Speaker 1>Welcome to Brainstuff from how stuff Works dot com where

0:00:03.840 --> 0:00:15.120
<v Speaker 1>smart Happens. Hi. I'm Marshall Brain with today's question, how

0:00:15.160 --> 0:00:19.479
<v Speaker 1>do self driving cars work? Who wouldn't like to have

0:00:19.520 --> 0:00:22.840
<v Speaker 1>a car that drives itself. The idea of being able

0:00:22.880 --> 0:00:26.240
<v Speaker 1>to sleep or to use your laptop during the commute

0:00:26.239 --> 0:00:29.600
<v Speaker 1>to work is certainly appealing. The idea of having your

0:00:29.640 --> 0:00:32.000
<v Speaker 1>car drop you off at the door of a store

0:00:32.560 --> 0:00:35.839
<v Speaker 1>and then go park itself is also attractive. And then

0:00:35.880 --> 0:00:39.520
<v Speaker 1>there's the safety thing. Human beings just are not that

0:00:39.600 --> 0:00:42.600
<v Speaker 1>good when it comes to driving, with millions of automobile

0:00:42.640 --> 0:00:46.760
<v Speaker 1>accidents causing damage, injury, and death every year. If we

0:00:46.840 --> 0:00:50.320
<v Speaker 1>all had self driving cars, it would be great. Two

0:00:50.360 --> 0:00:54.200
<v Speaker 1>recent announcements have made self driving cars seem a lot

0:00:54.240 --> 0:00:57.720
<v Speaker 1>closer to reality. The first comes from the Army, which

0:00:57.760 --> 0:01:01.200
<v Speaker 1>is now using self driving vehicle to guard a large

0:01:01.240 --> 0:01:04.920
<v Speaker 1>military facility and nuclear waste dump in Nevada. And then

0:01:05.000 --> 0:01:08.959
<v Speaker 1>there's Google, which recently announced that it's developed self driving

0:01:09.000 --> 0:01:12.319
<v Speaker 1>cars that have logged over a hundred thousand miles on

0:01:12.440 --> 0:01:16.759
<v Speaker 1>real roads in real traffic. What's going on inside these

0:01:16.800 --> 0:01:19.640
<v Speaker 1>cars to make it possible for them to drive themselves.

0:01:20.040 --> 0:01:22.240
<v Speaker 1>We can take a look at it. By looking at

0:01:22.400 --> 0:01:26.200
<v Speaker 1>Google cars or some of the cars that have participated

0:01:26.280 --> 0:01:31.280
<v Speaker 1>in DARPST Grand Challenge, you could, in theory, turn any

0:01:31.319 --> 0:01:34.320
<v Speaker 1>car into a self driving car. The first thing you

0:01:34.319 --> 0:01:36.880
<v Speaker 1>would have to do is make it possible for a

0:01:36.920 --> 0:01:40.280
<v Speaker 1>computer to control the car. This is not quite as

0:01:40.319 --> 0:01:43.920
<v Speaker 1>easy as it sounds, but it's relatively straightforward. The computer

0:01:44.000 --> 0:01:46.720
<v Speaker 1>needs to be able to turn the steering wheel, push

0:01:46.760 --> 0:01:50.200
<v Speaker 1>the accelerator and brake pedals, move the gear shift control,

0:01:50.320 --> 0:01:54.480
<v Speaker 1>and start the engine. These tasks, at least in experimental

0:01:54.520 --> 0:01:58.640
<v Speaker 1>self driving cars, are often accomplished with electric motors. A

0:01:58.680 --> 0:02:01.800
<v Speaker 1>motor is mounted so it can turn the existing steering

0:02:01.800 --> 0:02:05.240
<v Speaker 1>wheel for example. Another motor is able to put pressure

0:02:05.280 --> 0:02:08.519
<v Speaker 1>on the accelerator and brake pedals and so on. It's

0:02:08.560 --> 0:02:11.880
<v Speaker 1>not a particularly elegant solution, but it gets the job done.

0:02:12.280 --> 0:02:16.639
<v Speaker 1>As production cars become more advanced, drive by wire systems

0:02:16.639 --> 0:02:20.040
<v Speaker 1>are becoming a lot more common, So in a like

0:02:20.080 --> 0:02:23.840
<v Speaker 1>a Prius a Toyota Prius, if you push the brake pedal,

0:02:24.280 --> 0:02:27.079
<v Speaker 1>it might go to a computer and the computer actually

0:02:27.320 --> 0:02:31.960
<v Speaker 1>activates the brakes. There's no direct mechanical linkage between the

0:02:32.000 --> 0:02:36.000
<v Speaker 1>brake pedal and the brake mechanism that's called drive by

0:02:36.160 --> 0:02:38.720
<v Speaker 1>wire when the computer is sitting there in the middle.

0:02:39.320 --> 0:02:42.880
<v Speaker 1>In cars with drive by wire, a computer can hook

0:02:43.000 --> 0:02:46.440
<v Speaker 1>directly into that existing control system without having to have

0:02:46.560 --> 0:02:49.840
<v Speaker 1>these motors hanging out in the driver's area. The next

0:02:49.840 --> 0:02:52.960
<v Speaker 1>thing that a self driving car needs is sensors, and

0:02:53.040 --> 0:02:57.240
<v Speaker 1>here things get pretty complicated. Today, a human being drives

0:02:57.320 --> 0:03:00.440
<v Speaker 1>his car by using his or her eye is as

0:03:00.520 --> 0:03:04.440
<v Speaker 1>the soul sensor. The human visual system is amazing in

0:03:04.480 --> 0:03:07.959
<v Speaker 1>this regard because it's able to accurately judge the presence

0:03:07.960 --> 0:03:12.760
<v Speaker 1>of obstacles, their distances, their relative size based on distance,

0:03:12.800 --> 0:03:16.800
<v Speaker 1>their speed, and so on. A human being also recognizes

0:03:16.880 --> 0:03:19.560
<v Speaker 1>what he or she is seeing, So if a human

0:03:19.600 --> 0:03:23.520
<v Speaker 1>being sees a fence, you can accurately predict with high

0:03:23.560 --> 0:03:26.079
<v Speaker 1>certainty that that fence is not going to jump out

0:03:26.080 --> 0:03:28.200
<v Speaker 1>into the middle of the road and do something to you.

0:03:28.520 --> 0:03:31.000
<v Speaker 1>On the other hand, if there's a child playing with

0:03:31.040 --> 0:03:33.760
<v Speaker 1>a ball on a sidewalk, it's a completely different story,

0:03:33.840 --> 0:03:38.600
<v Speaker 1>and humans can recognize that instantly. Computer vision systems are

0:03:38.680 --> 0:03:42.000
<v Speaker 1>nowhere near this point in their development, so they rely

0:03:42.200 --> 0:03:46.800
<v Speaker 1>on extra sensors to provide more information. Self driving cars

0:03:46.880 --> 0:03:50.800
<v Speaker 1>do have camera based vision systems, and they use it

0:03:50.840 --> 0:03:55.040
<v Speaker 1>to see other cars, unexpected obstacles, road marking signs, and

0:03:55.080 --> 0:03:58.840
<v Speaker 1>so on. But in addition, self driving cars almost always

0:03:58.840 --> 0:04:01.440
<v Speaker 1>have GPS sense, or so they have a better idea

0:04:01.480 --> 0:04:04.760
<v Speaker 1>of exactly where they are and where they're pointing. They

0:04:04.800 --> 0:04:08.880
<v Speaker 1>also have light our systems laser scanners that can look

0:04:08.960 --> 0:04:12.880
<v Speaker 1>for nearby objects and accurately judge their distance. They also

0:04:13.000 --> 0:04:17.680
<v Speaker 1>have radar systems normally, and they may have infrared sensors

0:04:17.720 --> 0:04:21.719
<v Speaker 1>to improve night vision. These sensors all feed into a

0:04:21.760 --> 0:04:26.920
<v Speaker 1>powerful onboard computer, often multiple computers that are usually located

0:04:26.960 --> 0:04:29.599
<v Speaker 1>back in the back catch area of the trunk. They

0:04:29.680 --> 0:04:33.680
<v Speaker 1>could fill an entire trunk in fact, and these process

0:04:33.839 --> 0:04:37.279
<v Speaker 1>all the data that the sensors are gathering. The internal

0:04:37.279 --> 0:04:40.200
<v Speaker 1>computer will have access to a database of maps and

0:04:40.240 --> 0:04:44.880
<v Speaker 1>other relevant information as well. For example, engineers may pre

0:04:45.080 --> 0:04:50.880
<v Speaker 1>drive or route and pre catalog all the signs, road markings, curves, crosswalks,

0:04:50.920 --> 0:04:53.920
<v Speaker 1>traffic lights, and so on that the self driving car

0:04:54.000 --> 0:04:57.599
<v Speaker 1>will encounter along the road. This way, the car knows

0:04:57.680 --> 0:05:02.200
<v Speaker 1>what to expect and complan accordingly. Finally, the onboard computer

0:05:02.279 --> 0:05:05.960
<v Speaker 1>may also be communicating by radio, with bigger computers holding

0:05:06.040 --> 0:05:10.320
<v Speaker 1>even more data in nearby buildings. You might wonder why

0:05:10.360 --> 0:05:14.120
<v Speaker 1>the driving task, which seems effortless to just about any

0:05:14.160 --> 0:05:17.640
<v Speaker 1>normal human adult, would be so hard for a computer.

0:05:18.200 --> 0:05:22.160
<v Speaker 1>It's because the human brain is deceptive. Although it's small

0:05:22.200 --> 0:05:25.760
<v Speaker 1>and silent compared to a desktop computer, the human brain

0:05:25.960 --> 0:05:30.560
<v Speaker 1>still has more power than mankind's biggest fills a whole

0:05:30.640 --> 0:05:34.680
<v Speaker 1>building supercomputers. The massive amount of computing power in the

0:05:34.760 --> 0:05:37.560
<v Speaker 1>human brain is what makes driving appear to be so

0:05:37.640 --> 0:05:41.320
<v Speaker 1>easy for us. In thirty or forty years, our computers

0:05:41.320 --> 0:05:43.839
<v Speaker 1>will have that same kind of power in a small

0:05:43.920 --> 0:05:50.039
<v Speaker 1>package if things keep progressing at their current rate. Be

0:05:50.120 --> 0:05:52.760
<v Speaker 1>sure to check out our new video podcast, Stuff from

0:05:52.800 --> 0:05:55.640
<v Speaker 1>the Future. Join how Stuff Work staff as we explore

0:05:55.680 --> 0:06:01.080
<v Speaker 1>the most promising and perplexing possibilities of tomorrow. The Housuffork's

0:06:01.080 --> 0:06:04.359
<v Speaker 1>iPhone app has arrived. Download it today on iTunes.