WEBVTT - Search Engine Presents: Are you a good driver? 0:00:00.040 --> 0:00:02.480 Hey, there are odd lots listeners. I'm Tracy Alloway and 0:00:02.560 --> 0:00:03.000 I'm Jill. 0:00:03.040 --> 0:00:03.560 Why isn't though? 0:00:03.640 --> 0:00:06.159 And we want to welcome you to a special presentation 0:00:06.280 --> 0:00:09.720 of the podcast search Engine. We all know that artificial 0:00:09.760 --> 0:00:13.120 intelligence might replace all sorts of jobs humans do today, 0:00:13.360 --> 0:00:16.799 but for most of us that's still mostly theoretical. There's 0:00:16.840 --> 0:00:19.919 one job, though, where robots are already taking the wheel, 0:00:20.079 --> 0:00:21.320 and that is driving. 0:00:21.640 --> 0:00:23.360 In fact, it's one of the most common jobs in 0:00:23.400 --> 0:00:26.239 America for young men without college degrees, and over the 0:00:26.280 --> 0:00:29.520 course of two episodes, Search Engine tackles both the promise 0:00:29.600 --> 0:00:31.600 and the peril this growing technology. 0:00:31.840 --> 0:00:34.320 In part one, Are You a Good Driver? The Search 0:00:34.360 --> 0:00:36.920 Engine team tells the story of how a small secret 0:00:36.960 --> 0:00:39.879 team at Google spent fifteen years teaching a computer to 0:00:39.960 --> 0:00:43.000 drive from a failed robot in the Mojave Desert to 0:00:43.080 --> 0:00:45.480 a vehicle that might actually be the safest on the road. 0:00:45.800 --> 0:00:49.680 This episode tracks the engineering breakthroughs the nearer catastrophes, and 0:00:49.760 --> 0:00:52.720 takes a skeptical look at the safety data behind Weaymo's 0:00:52.760 --> 0:00:55.520 claim that its cars are ninety percent safer than human 0:00:55.640 --> 0:00:57.160 drivers in serious crashes. 0:00:57.560 --> 0:01:00.200 All of it boils down to one big question. All 0:01:00.200 --> 0:01:03.000 the robots actually say for drivers than we are. Enjoy 0:01:03.040 --> 0:01:05.680 this presentation of Search Engine, and be sure to catch 0:01:05.760 --> 0:01:09.280 part two, titled The Trial of the Driverless Car, available 0:01:09.280 --> 0:01:33.480 wherever you get your podcasts. 0:01:37.800 --> 0:01:40.119 Before we start the story today, I want to ask 0:01:40.160 --> 0:01:43.600 you to imagine a different version of your life. You're you, 0:01:44.080 --> 0:01:47.200 but it's almost two hundred years ago, and unfortunately and 0:01:47.280 --> 0:01:51.360 our hypothetical, it's Monday morning. It's Monday morning, and it's 0:01:51.480 --> 0:01:55.440 very early pre dawn. You wake up to this really 0:01:55.480 --> 0:01:59.200 hard wrapping at your window. That's the knocker Upper here 0:01:59.240 --> 0:02:01.360 to get you up for where we're in the eighteen 0:02:01.440 --> 0:02:05.040 hundreds before the invention of the adjustable alarm clock. The 0:02:05.080 --> 0:02:08.840 knocker Upper is a job. The knocker Upper walks the 0:02:08.840 --> 0:02:11.679 neighborhood with a long stick and taps it on the 0:02:11.720 --> 0:02:14.240 windows of people's houses early in the morning to wake 0:02:14.280 --> 0:02:17.440 them up for work. Who wakes up the locker rupper 0:02:17.440 --> 0:02:20.720 for work? Nobody knows. But this is a job, a 0:02:20.840 --> 0:02:25.280 job that'll actually exist for another century. Outside the gas 0:02:25.280 --> 0:02:28.000 street lamps are still burning. The lamplighter lit them the 0:02:28.080 --> 0:02:31.359 night before. He's supposed to come at dawn to extinguish them, 0:02:31.440 --> 0:02:34.320 but it's so early that he has it yet. Your 0:02:34.400 --> 0:02:36.399 lamplighter is one of those neighbors. You have a deep 0:02:36.440 --> 0:02:40.360 fondness for a fixture. Every day you watch him make 0:02:40.400 --> 0:02:42.919 the rounds at dusk with his ladder and his light. 0:02:44.639 --> 0:02:48.840 You yourself are a driver. Professional driver two hundred years 0:02:48.840 --> 0:02:52.000 ago is also a job. You're a person who sits 0:02:52.000 --> 0:02:54.160 on a coach and holds the reins of a horse. 0:02:54.880 --> 0:02:57.760 You take passengers where they want to go. You start 0:02:57.760 --> 0:03:05.840 your workday. Okay, hypothetical. Over two of those jobs are 0:03:05.880 --> 0:03:09.360 obviously so long disappeared that most people don't know about them. 0:03:09.760 --> 0:03:13.040 The knocker upper is your iPhone alarm. The lamplighter is 0:03:13.040 --> 0:03:17.400 the electric street light. The third one driver has persisted 0:03:18.080 --> 0:03:20.840 as a job for some, as a routine human task 0:03:20.960 --> 0:03:24.600 for nearly everyone else. This is a story about whether 0:03:24.639 --> 0:03:27.880 that's about to change. It's about how the word driver, 0:03:28.240 --> 0:03:31.040 which right now makes me picture a human, could soon 0:03:31.120 --> 0:03:34.359 transform to refer to a machine, the same way the 0:03:34.400 --> 0:03:39.840 words dishwasher, printer, and computer all did. I've thought about this, 0:03:40.080 --> 0:03:42.800 maybe too much, in the year I've been working on 0:03:42.840 --> 0:03:46.600 this story. In conversations constantly, I'd asked the humans I 0:03:46.680 --> 0:03:50.920 meant the same question, Are you a good driver? Are 0:03:51.040 --> 0:03:52.760 you do you consider yourself a good driver. 0:03:54.040 --> 0:03:59.440 I do within limits. I think I'm a good driver 0:03:59.560 --> 0:04:02.680 because I understand the limitations of my driving. 0:04:03.320 --> 0:04:06.680 This is Alex Davies. He wrote an excellent book called Driven, 0:04:06.880 --> 0:04:10.160 the Race to create the Autonomous Car. Alex, like me, 0:04:10.280 --> 0:04:13.720 thinks a lot about human driving about his own personal limitations. 0:04:14.240 --> 0:04:15.280 What are the limitations? 0:04:15.840 --> 0:04:18.880 The limitations are that I can't always pay attention to 0:04:18.920 --> 0:04:23.240 everything that I get tired. I've been trying really hard 0:04:23.839 --> 0:04:27.800 to be calmer in the road. My husband and I 0:04:27.839 --> 0:04:31.920 are expecting our first baby this fall. Congratulations, thank you, 0:04:32.240 --> 0:04:35.000 and I thought that, along with reading all the baby books, 0:04:35.040 --> 0:04:37.679 a good project to work on is just be calmer 0:04:37.720 --> 0:04:38.680 in the car. 0:04:39.560 --> 0:04:42.200 A very good resolution, because, of course, for most of us, 0:04:42.440 --> 0:04:46.320 driving is the riskiest behavior we routinely engage in. In fact, 0:04:46.400 --> 0:04:49.359 even Alex, despite his good intentions, would actually get in 0:04:49.400 --> 0:04:51.559 a car accident just a few months after we first spoke. 0:04:52.440 --> 0:04:54.400 He was okay, it was the car that was totaled. 0:04:56.040 --> 0:04:58.719 Safety is the entire pitch for the driver of this car, 0:04:58.880 --> 0:05:01.800 which is really a car and by a computer driver. 0:05:01.880 --> 0:05:05.159 LESE cars don't get drunk, tired, or distracted. They never 0:05:05.279 --> 0:05:09.160 text or feel road rage, and these drivers cars, they 0:05:09.200 --> 0:05:13.719 aren't the future. They're actually already here. But it's funny. 0:05:13.720 --> 0:05:15.479 If you just don't happen to live in a place 0:05:15.520 --> 0:05:18.320 that already has them, it's easy to not see how 0:05:18.360 --> 0:05:22.719 fast things are changing. Robo taxis like Waimo are operating 0:05:22.720 --> 0:05:25.880 in ten American cities, providing millions of rides to Americans. 0:05:26.880 --> 0:05:29.440 In China, the rollout is happening even more widely. They're 0:05:29.440 --> 0:05:32.360 in twice as many cities. But here, if you live 0:05:32.400 --> 0:05:35.240 in a place like San Francisco or Austin, today, a 0:05:35.320 --> 0:05:38.680 driver's car is about as exotic as an uber. A 0:05:38.720 --> 0:05:41.720 passenger in those cities opens up their phone and decides 0:05:41.760 --> 0:05:45.159 who should drive them, a human driver or a robot driver. 0:05:48.040 --> 0:05:51.360 How that happened is a story, a story we are 0:05:51.360 --> 0:05:54.599 living through right now, whose ending promise is to totally 0:05:54.600 --> 0:05:57.400 reshape the places we live. And today we're going to 0:05:57.400 --> 0:06:03.760 tell you how we got here. In chapters, Chapter one 0:06:04.360 --> 0:06:09.080 dreams without drivers. So it turns out this dream that 0:06:09.160 --> 0:06:12.120 inventors have had to replace the human driver with some 0:06:12.200 --> 0:06:14.760 kind of machine. That dream is about as old as 0:06:14.760 --> 0:06:15.600 the Lamplighters. 0:06:16.480 --> 0:06:20.080 People have been thinking about a self driving car for 0:06:21.640 --> 0:06:23.880 so it's about as long as there's been a human 0:06:24.080 --> 0:06:28.240 driven car. Why, there's this funny thing you lose when 0:06:28.279 --> 0:06:30.760 you move from the horse to the human driven car, 0:06:30.800 --> 0:06:34.640 which is that in a horse drawn carriage, the horse 0:06:34.720 --> 0:06:36.280 is not just going to run off a cliff. If 0:06:36.320 --> 0:06:40.800 you let go of the reins, you lose sentience in 0:06:40.839 --> 0:06:41.520 your vehicle. 0:06:45.640 --> 0:06:50.320 When automobiles first arrived, these powerful and nonsensient cars, there's 0:06:50.320 --> 0:06:52.960 actually a passionate fight to keep them off the streets. 0:06:53.760 --> 0:06:56.800 It was the eighteen hundreds, and people feared these new things, 0:06:57.640 --> 0:07:00.960 the steam powered vehicles thundering down the roads that soon 0:07:00.960 --> 0:07:04.599 evolved into gas powered vehicles, also thundering down the roads. 0:07:05.880 --> 0:07:09.039 The fear was partly about jobs. These vehicles were seen 0:07:09.080 --> 0:07:11.800 as a huge threat to a whole network of working 0:07:11.880 --> 0:07:16.280 class jobs. Horse breeders and horse farriers, horse feed suppliers, 0:07:16.400 --> 0:07:20.640 horse manure haulers, horse carriage manufacturers. Not to mention the 0:07:20.680 --> 0:07:24.680 teamsters Teamsters today the word makes me think of the 0:07:24.680 --> 0:07:28.160 Teamsters union, But originally the teamsters were the workers who 0:07:28.240 --> 0:07:31.800 drove teams of horses. Teamsters were like truckers before we 0:07:31.800 --> 0:07:36.920 had trucks. Cars seemed to imperil all these horse related jobs. 0:07:37.360 --> 0:07:40.040 And even if you weren't worried about these workers, the 0:07:40.080 --> 0:07:43.760 cars were also less safe. Some anti car activists battled 0:07:43.760 --> 0:07:47.920 to stop or slow the new technology, mainly with regulations. 0:07:48.800 --> 0:07:51.120 There were red flag laws, which said if you had 0:07:51.120 --> 0:07:53.760 an automobile, you had to hire a person to walk 0:07:53.800 --> 0:07:56.400 in front of it, waving a giant red flag to 0:07:56.480 --> 0:08:01.720 warn people. In Pennsylvania, a law was prepared requiring horseless 0:08:01.760 --> 0:08:06.000 carriage drivers who encountered livestock to stop, disassemble their car, 0:08:06.520 --> 0:08:10.160 and hide the parts behind the bushes. The governor vetoed it. 0:08:12.920 --> 0:08:16.600 But to think about these crazy anti car activists is 0:08:16.720 --> 0:08:20.320 that directionally they were right. Those cars did initially wipe 0:08:20.320 --> 0:08:22.480 out a lot of jobs, even if they created more, 0:08:23.040 --> 0:08:26.040 and cars were very unsafe. The cities that threw their 0:08:26.080 --> 0:08:30.240 doors open to cars without regulation were rewarded with astonishing 0:08:30.280 --> 0:08:34.160 death rates. Detroit let drivers pretty much run wild. In 0:08:34.200 --> 0:08:38.000 the early nineteen hundreds, deaths accumulated in a Detroit without 0:08:38.080 --> 0:08:41.760 drivers licenses, stop lights, or turn signals. Many of those 0:08:41.800 --> 0:08:46.040 deaths were children. It took decades for society to mostly 0:08:46.120 --> 0:08:49.000 learn to live with cars. The rest of the story 0:08:49.120 --> 0:08:52.520 is just the world you grew up in. We invented laws, licenses, 0:08:52.600 --> 0:08:55.760 drivers ed, We learned to better design roads. We invented 0:08:55.760 --> 0:08:59.280 the highway, the seatbelt, the airbag. All those things made 0:08:59.320 --> 0:09:04.800 driving less debts, although the smartphone reverse some of that progress. Nationally, Today, 0:09:05.000 --> 0:09:07.559 deaths from cars are about as common in America as 0:09:07.600 --> 0:09:10.720 deaths from guns or opioids, about one in one hundred. 0:09:11.840 --> 0:09:14.120 It'll probably happen to someone you know in your life, 0:09:14.720 --> 0:09:18.280 maybe several someone's. Whether or not you see that as 0:09:18.320 --> 0:09:22.240 an urgent problem to solve depends on you. But as 0:09:22.360 --> 0:09:24.440 long as there have been cars, there have been people 0:09:24.480 --> 0:09:27.040 who wanted to truly solve what's left of the safety 0:09:27.040 --> 0:09:30.160 problem the best way we knew how. They wanted to 0:09:30.160 --> 0:09:33.439 make the car more like the horse it replaced, make 0:09:33.480 --> 0:09:34.520 the car more sentient. 0:09:36.000 --> 0:09:39.800 So that thought is there early and like early visions 0:09:39.840 --> 0:09:44.480 have hit include, oh well, we'll have radio controlled cars, 0:09:44.920 --> 0:09:49.000 because they had radios at the time. There's a real 0:09:49.120 --> 0:09:52.800 effort at one point to build magnets under the road, 0:09:53.840 --> 0:09:57.440 and at each stage what a self driving car can 0:09:57.520 --> 0:10:01.400 be is dictated by the technology that's available at the time. 0:10:01.880 --> 0:10:05.439 For the most part, no one's thinking that much about 0:10:05.880 --> 0:10:09.680 a vehicle that thinks for itself. They're just thinking about 0:10:09.720 --> 0:10:12.640 a vehicle that the person in it doesn't have to drive. 0:10:13.880 --> 0:10:18.079 Many different attempts, many different failures, as many wonders as 0:10:18.080 --> 0:10:22.120 we invented, we could not approach nature's most majestic creation, 0:10:23.200 --> 0:10:26.280 a horse's brain, at least not until the turn of 0:10:26.280 --> 0:10:29.720 the millennium. 0:10:30.880 --> 0:10:33.880 Are to. 0:10:38.760 --> 0:10:41.360 Deep within the Department of Defense, there's a little known 0:10:41.400 --> 0:10:44.400 military agency that has created some of the most innovative 0:10:44.440 --> 0:10:49.000 technology of the twentieth century. This is the story of Dark. 0:10:49.040 --> 0:10:52.040 Chapter two, DARPA's Million Dollar Prize. 0:10:52.440 --> 0:10:56.880 DARPA's current goal is to develop autonomous military vehicles machines 0:10:56.920 --> 0:10:59.559 that can operate on their own without drivers. 0:11:00.040 --> 0:11:01.600 Darpe has always been intrigued with him. 0:11:01.720 --> 0:11:04.600 This is from a documentary called The Million Dollar Challenge. 0:11:04.720 --> 0:11:07.440 Honestly less a doc more an ad for DARPA, the 0:11:07.480 --> 0:11:11.160 Pentagon's research arm. DARPA's mission is to try to keep 0:11:11.160 --> 0:11:14.920 American technology one generation ahead of everybody else. It doesn't 0:11:14.920 --> 0:11:18.440 always work, but DARPA has invented or funded a lot 0:11:18.840 --> 0:11:22.280 GPS and the M sixteen, the early Internet, and the 0:11:22.280 --> 0:11:25.640 Predator drone. In two thousand and two, DARPA decided to 0:11:25.640 --> 0:11:28.720 pursue the driver's car in a very unusual way. 0:11:29.559 --> 0:11:31.680 The director of DARPA at the time, a guy named 0:11:31.679 --> 0:11:35.400 Tony Tether, who had been a door to door salesman 0:11:35.679 --> 0:11:39.120 in his use definitely has that flare in that way 0:11:39.160 --> 0:11:44.800 of thinking, says, let's have a contest. Let's see who 0:11:44.800 --> 0:11:49.000 can put all of these ingredients that we've developed together 0:11:49.320 --> 0:11:53.320 into a proper self driving car. His original idea is 0:11:53.440 --> 0:11:56.760 we'll drive him down the Las Vegas Strip that's almost 0:11:56.800 --> 0:11:58.880 immediately next because it's insane. 0:11:59.080 --> 0:12:03.000 Oh right, you would have to like literally gridlock a 0:12:03.160 --> 0:12:07.480 huge American city so people could put robot cars on 0:12:07.520 --> 0:12:08.679 it exactly. 0:12:09.000 --> 0:12:10.640 So he says, Okay, do you know what, We'll do 0:12:10.679 --> 0:12:12.720 it in the desert. We'll do it in the desert 0:12:12.800 --> 0:12:17.520 outside Las Vegas, and anyone who wants to can make 0:12:17.559 --> 0:12:20.320 a team build a self driving car, bring it to 0:12:20.400 --> 0:12:22.079 the desert, and we'll race them. 0:12:22.600 --> 0:12:25.800 The driver that DARPA wanted to replace was the American soldier. 0:12:26.400 --> 0:12:29.240 DARPA wanted a vehicle that could drive itself down roads 0:12:29.280 --> 0:12:32.680 that might be filled with hidden explosive devices. So, in 0:12:32.720 --> 0:12:35.200 this moment at the tail end of the dot com boom, 0:12:35.520 --> 0:12:38.559 darpest trying to inspire tech to build something besides another 0:12:38.600 --> 0:12:42.600 website darpest. Tony Tether announces that the prize for whoever 0:12:42.640 --> 0:12:45.640 can win its Grand Challenge will be one million dollars. 0:12:46.520 --> 0:12:50.320 The rules for very open there were little rules like 0:12:50.360 --> 0:12:53.600 you couldn't have two vehicles communicating with one another, but 0:12:53.720 --> 0:12:56.320 you could build any kind of vehicle you wanted, could 0:12:56.320 --> 0:12:58.040 have six wheels. It could be a truck, it could 0:12:58.080 --> 0:13:01.640 be a motorcycle, could be a trice. It just couldn't 0:13:01.679 --> 0:13:04.640 attack other vehicles. That was rolled out early on. 0:13:04.920 --> 0:13:06.800 Oh, was that a concern that people would just like 0:13:07.360 --> 0:13:09.880 sort of battlebot the thing you're auto's vehicle would have 0:13:09.960 --> 0:13:11.880 like a little shredder that would take out somebody else's. 0:13:12.800 --> 0:13:15.520 Someone asked in the first Q and A at this 0:13:15.720 --> 0:13:18.520 like they said, can we attack other vehicles? And they 0:13:18.520 --> 0:13:21.760 said no. And it's funny you bring up BattleBots because 0:13:21.880 --> 0:13:26.319 a lot of teams who entered this had BattleBots history interesting. 0:13:26.480 --> 0:13:30.920 They were used to building robots for interesting purposes, and 0:13:30.960 --> 0:13:33.280 when they caught wind of this, they said, we can 0:13:33.360 --> 0:13:36.360 do this. We can scrap together some money and this 0:13:36.400 --> 0:13:37.320 will just be fun. 0:13:41.120 --> 0:13:43.920 I'm going to tell you what happened in this robot 0:13:44.000 --> 0:13:46.760 race in the desert, not because I care so much 0:13:46.760 --> 0:13:49.880 about these early robot vehicles, but because I care a 0:13:49.920 --> 0:13:53.240 lot about the engineers who were making them. These would 0:13:53.280 --> 0:13:55.000 be the people who would later go on to lead 0:13:55.000 --> 0:13:58.640 development for the billion dollar companies creating today's drive those cars. 0:13:59.840 --> 0:14:02.840 The people had very different views about how to get 0:14:02.840 --> 0:14:06.080 that technology ready, different values when it came to things 0:14:06.160 --> 0:14:10.920 like the acceptability of risking human life, abstract differences that 0:14:10.920 --> 0:14:14.000 would become very concrete later on, to the point where 0:14:14.000 --> 0:14:19.440 people would be charged with federal crimes. That's the future. 0:14:20.040 --> 0:14:22.080 But listening to this part of the story, what I 0:14:22.240 --> 0:14:25.080 listen for is how much of it can you detect already? 0:14:25.320 --> 0:14:28.960 How much of the differences already present. The first engineer 0:14:28.960 --> 0:14:30.480 I want you to pay attention to is a man 0:14:30.560 --> 0:14:35.360 named Chris Armson, and way back in two thousand and two, 0:14:35.680 --> 0:14:37.680 how did you end up being part of the Dark 0:14:37.680 --> 0:14:38.600 Program challenge? 0:14:39.960 --> 0:14:40.880 It sounded like fun. 0:14:43.080 --> 0:14:45.800 Chris these days, the CEO of a large tech company 0:14:46.200 --> 0:14:50.200 back then, a PhD student at Carnegie Mellon University. When 0:14:50.200 --> 0:14:52.200 he first got recruited for the race, he was out 0:14:52.200 --> 0:14:55.560 in the field observing a robot as it crept across 0:14:55.560 --> 0:14:58.720 the Autacama Desert training for its future deployment on the 0:14:58.720 --> 0:14:59.560 surface of Mars. 0:15:00.040 --> 0:15:03.200 H advisor came down and was really excited about this 0:15:03.360 --> 0:15:06.240 Darker Grind challenge thing, and the idea that you'd have 0:15:06.320 --> 0:15:09.880 a robot run across the desert at fifty miles an 0:15:09.880 --> 0:15:14.560 hour just sounded exciting having spent the last couple of 0:15:14.560 --> 0:15:17.960 weeks walking behind a robot at very low speed. 0:15:20.080 --> 0:15:22.920 So Chris would join Carnegie Mellon's Red Team and help 0:15:22.960 --> 0:15:25.920 build a car called Sandstorm, a bright red humvey with 0:15:25.960 --> 0:15:29.640 the top lopped off, a plethora of futuristic sensors mounted 0:15:29.640 --> 0:15:32.560 to it like scanners a crackpot would use to search 0:15:32.560 --> 0:15:35.960 for aliens. You can see Chris back in that documentary. 0:15:36.280 --> 0:15:38.200 He explains to the filmmaker at the time that the 0:15:38.200 --> 0:15:40.560 hard part, of course, isn't the vehicle, it's the driver. 0:15:41.280 --> 0:15:43.480 How do you even begin to teach a computer to 0:15:43.480 --> 0:15:45.320 operate a hum vy at all? How does a. 0:15:45.320 --> 0:15:46.720 Computer make the steering wheel turn? 0:15:46.760 --> 0:15:48.920 How does a computer change the. 0:15:48.640 --> 0:15:50.720 Pressure on the break and the throttle? Those are the 0:15:50.760 --> 0:15:52.600 issues that we're fighting through right now. 0:15:53.560 --> 0:15:56.880 The answer Sandstorm represented the best entry from the contest's 0:15:56.920 --> 0:16:00.480 traditional academic crowd, but there's a different crowd there too. 0:16:00.960 --> 0:16:04.600 Represented best by a man named Anthony Lewandowski. Can you 0:16:04.600 --> 0:16:06.120 tell me about Anthony Lewandowski? 0:16:07.000 --> 0:16:18.600 Anthony Lewandowski. Where to begin? So Anthony is like an entrepreneur. 0:16:19.040 --> 0:16:24.080 He's a really charming guy. He's six foot six, He's 0:16:24.280 --> 0:16:28.480 gangly as all get down. He grew up mostly in 0:16:28.560 --> 0:16:32.360 Belgium because his mom was working for the EU. For 0:16:32.520 --> 0:16:36.240 high school, he moved to Marin to live with his dad. 0:16:37.280 --> 0:16:38.760 And he's a hustler. 0:16:39.400 --> 0:16:41.040 My name is Anthony Lewandowski. 0:16:42.400 --> 0:16:44.320 I was a grad student at Berkeley. 0:16:44.480 --> 0:16:47.720 Instead of continuing on to finish my PhD, I decided 0:16:47.760 --> 0:16:49.960 it was much better to do the Grand Challenge. 0:16:50.440 --> 0:16:52.960 We asked Anthony for an interview. He didn't respond, but 0:16:53.040 --> 0:16:55.600 here he is in the footage from back then. Anthony 0:16:55.640 --> 0:16:58.640 did not have the engineering experience or resources of a 0:16:58.640 --> 0:17:01.880 team like Carnegie Mellons Red Team. So you tried something 0:17:01.960 --> 0:17:04.879 very different, a vehicle that had almost no chance of 0:17:04.920 --> 0:17:07.840 winning the race, but which was also perfectly designed to 0:17:07.880 --> 0:17:10.359 stand out to get him a lot of attention, maybe 0:17:10.359 --> 0:17:14.840 a job. The race's only self driving motorcycle, it was 0:17:14.920 --> 0:17:18.240 named ghost Rider, a stubby little thing covered in stickers 0:17:18.440 --> 0:17:20.639 with then inten on the back and cameras on the front. 0:17:21.840 --> 0:17:24.639 There's a steering actuator on the top here, which allows 0:17:24.680 --> 0:17:29.040 us to modify the steering angle. So basically, if you're driving, 0:17:29.080 --> 0:17:31.480 you start to follow the left, you steer left. That 0:17:31.560 --> 0:17:33.159 makes you turn the left, and then you get the 0:17:33.160 --> 0:17:35.120 tripleal acceleration to put you back up to the right. 0:17:35.720 --> 0:17:38.520 And you're monitoring that in real time and making small adjustments, 0:17:38.520 --> 0:17:39.720 and you stay bounced. 0:17:42.320 --> 0:17:45.919 Stroll Blight is on. The command from the tower is 0:17:45.960 --> 0:17:49.399 to move, ladies and gentlemen Sandstorm. 0:17:50.240 --> 0:17:52.440 The race happens on a Saturday in March of two 0:17:52.440 --> 0:17:52.840 thousand and. 0:17:52.840 --> 0:17:59.879 Four, autonomous vehicle traversing the desert with the goal of 0:18:00.119 --> 0:18:07.680 keeping our young military personnel out of harm's way. 0:18:08.600 --> 0:18:11.840 Oh yeah, what happens the first time they try to 0:18:11.840 --> 0:18:12.560 do this competition? 0:18:14.000 --> 0:18:16.920 The two thousand and four Grand Challenge is an utter 0:18:17.760 --> 0:18:19.520 hysterical disaster. 0:18:22.280 --> 0:18:27.320 Disaster Number one ghost Rider the motorcycle Anthony Lewandowski forgot 0:18:27.320 --> 0:18:30.199 to flip on the switch for the stabilization system. The 0:18:30.240 --> 0:18:34.560 bike immediately topples ghost Rider down. 0:18:34.640 --> 0:18:43.119 Anthony good effort, and then every vehicle after it fails miserably. 0:18:43.480 --> 0:18:46.800 Like one vehicle drives up onto a burm flips off. 0:18:47.160 --> 0:18:50.920 One vehicle, drives straight out, does an inexplicable U turn 0:18:51.440 --> 0:18:54.040 and just drives back to the starting line. And the 0:18:54.119 --> 0:18:57.000 rules are that once your vehicle starts, you can't do anything. 0:18:57.800 --> 0:19:01.080 Even Sandstorm got stuck on a burm. Chris Urmson just 0:19:01.240 --> 0:19:03.200 standing there, unable to help his robot. 0:19:03.720 --> 0:19:06.000 Poor thing was trying to get going, but its wheels 0:19:06.000 --> 0:19:10.000 were just spinning on the gravel and tried so hard 0:19:10.080 --> 0:19:12.200 that it actually melted the rubber of the tires. 0:19:12.200 --> 0:19:13.480 And so there's this plums of. 0:19:13.440 --> 0:19:15.800 Black squoke before they killed it. 0:19:16.600 --> 0:19:20.320 For the roboticists, this was obviously very disappointing. Chris Urmson 0:19:20.359 --> 0:19:22.919 compared it to an Olympic marathon where the best runner 0:19:22.960 --> 0:19:26.200 only makes it two of the twenty six miles. What 0:19:26.240 --> 0:19:29.240 this contest had done, though, was it had flushed all 0:19:29.280 --> 0:19:32.040 these inventors out. It had jumpstarted the scene that would 0:19:32.040 --> 0:19:35.280 develop this technology. One of the most important people there 0:19:35.320 --> 0:19:39.240 that day, actually just watching, was someone I haven't mentioned yet, 0:19:39.480 --> 0:19:42.159 a legendary roboticist named Sebastian Thrun. 0:19:43.080 --> 0:19:46.600 Sebastian Thrun, he was at the first Grand Challenge. He 0:19:46.640 --> 0:19:50.439 didn't bring a team, he wasn't participating. DARPA wanted to 0:19:50.440 --> 0:19:52.959 show off some other projects they'd been funding, including one 0:19:53.000 --> 0:19:55.479 of his robots. So he brings the robot and so 0:19:55.560 --> 0:20:00.159 he's there and he watches this disaster or anything's can 0:20:00.200 --> 0:20:01.040 do better for mess. 0:20:03.440 --> 0:20:06.160 I looked at the very first iteration of this quan challenge, 0:20:06.200 --> 0:20:08.119 but it didn't participate. It was a spectator. 0:20:08.960 --> 0:20:11.119 This, of course is Sebastian Thrun. He grew up in 0:20:11.160 --> 0:20:14.160 West Germany, moved to the US Toddic Carnegie Mellon before 0:20:14.160 --> 0:20:18.720 moving to Stanford. Watching that day, he saw this fundamental error. 0:20:18.760 --> 0:20:20.560 He believed all the entrance had made. 0:20:21.200 --> 0:20:24.560 I saw that all the teams treated this like a 0:20:24.560 --> 0:20:26.919 hardware problem. They looked at this and say, we have 0:20:27.000 --> 0:20:31.360 to build a bigger wheels and bigger chassis and so on. 0:20:32.480 --> 0:20:34.680 And I looked at this and said, about wait a minute. 0:20:34.960 --> 0:20:37.400 The challenge really is to build a self driving car. 0:20:37.640 --> 0:20:40.440 They can drive for the desert. I can get a 0:20:40.480 --> 0:20:43.080 rental car. They can do it just fine, provided as 0:20:43.080 --> 0:20:45.719 a person insight and the challenges we need to take 0:20:45.760 --> 0:20:47.560 the person out of the driver's seat and replace it 0:20:47.640 --> 0:20:51.000 by computer. That is not a problem with bigger tires. 0:20:51.080 --> 0:20:55.480 That's actually be a software problem. 0:20:55.640 --> 0:21:00.000 Sebastian Thrun had a dual background robotics and artificial intelligence, 0:21:00.480 --> 0:21:03.800 which probably explains his focus here on the robot driver's mind. 0:21:04.240 --> 0:21:07.600 He was thinking about something else too. The military wanted 0:21:07.600 --> 0:21:10.400 this tech to replace a relatively small number of drivers 0:21:10.400 --> 0:21:14.400 in its war zones, but Sebastian was already imagining something bigger. 0:21:15.119 --> 0:21:18.440 What would happen to traffic deaths worldwide? If one day 0:21:18.680 --> 0:21:20.600 everyone had access to a driverless car. 0:21:21.040 --> 0:21:23.399 I had experiences of losing people in my life to 0:21:23.480 --> 0:21:26.359 traffic accidents, and I felt we lost over the million 0:21:26.359 --> 0:21:29.040 people in the world to traffic accidents. Wouldn't it be 0:21:29.080 --> 0:21:32.480 amazing if Dabok invented something that would save a million 0:21:32.520 --> 0:21:33.120 lives a year. 0:21:33.920 --> 0:21:37.280 In October of two thousand and five, forty three teams 0:21:37.320 --> 0:21:40.439 have brought their vehicles to compete in a unique event, 0:21:41.200 --> 0:21:44.840 a race driven not by testosterone but computer. 0:21:44.520 --> 0:21:50.480 Coke Chapter three Machine Learning. 0:21:54.320 --> 0:21:57.640 The race course is a circular maze that zigzags for one 0:21:57.680 --> 0:21:58.359 hundred and thirty two. 0:21:58.440 --> 0:22:02.080 Eighteen months later this second Grand Challenge, DARPA doubled the 0:22:02.080 --> 0:22:05.280 bounty two million dollars. This footage is from a PBS 0:22:05.320 --> 0:22:08.720 documentary called The Great Robot Race, narrated to My Mild 0:22:08.840 --> 0:22:13.160 Joy by John Lithgow. Familiar faces have returned. Chris Earmsen 0:22:13.240 --> 0:22:16.119 back with the Carnegie Mellon team, the Sign with two vehicles, 0:22:16.400 --> 0:22:21.200 Highlander and Sandstorm. Anthony Lewandowski back with his motorcycle, which 0:22:21.320 --> 0:22:24.280 still doesn't work. He's knocked out in the qualifiers. And 0:22:24.320 --> 0:22:28.640 now there's also Stanford's entrant compared to Sandstorm, the bulked 0:22:28.680 --> 0:22:32.760 up hummer. The car looks easily a blue suv donated 0:22:32.760 --> 0:22:36.399 by Volkswagen. A baby face, Run smiles next to his 0:22:36.440 --> 0:22:37.760 soccer mom looking vehicle. 0:22:38.359 --> 0:22:41.280 The Vika's name is Stanley, so Stanley is nothing else 0:22:41.320 --> 0:22:44.360 but Stanford, but it also gives the vehicle a personality. 0:22:45.440 --> 0:22:47.080 If you think of the Vega more and more as 0:22:47.119 --> 0:22:48.600 an intelligent decision maker. 0:22:50.440 --> 0:22:54.600 Run is a computer scientist, and Thrun really broad more 0:22:54.720 --> 0:22:57.879 artificial intelligence, which at the time we're talking two thousand 0:22:57.920 --> 0:23:02.520 and five was still rather primitive, especially compared to what 0:23:02.560 --> 0:23:06.000 we have today. But he could use it to teach 0:23:06.480 --> 0:23:10.320 his vehicle how to recognize the road and how to 0:23:10.359 --> 0:23:13.640 do it much faster. They found a dirt road out 0:23:13.640 --> 0:23:15.920 near Stanford, and they drive it down a dirt road 0:23:16.960 --> 0:23:20.000 and have the car's cameras record what they were seeing. 0:23:20.760 --> 0:23:24.040