October 27, 2017

Driverless Cars 1: The Grand Challenge

by StartUp

Background show artwork for StartUp

The Story In 2004, the U.S. government held a race unlike anything that had come before it. It was called the DARPA Grand Challenge, and it followed a 150-mile route through the rugged Mojave Desert. The participants were cars, trucks, ATVs, and one motorcycle. The catch? Each vehicle was required to drive itself—no remote control, no human intervention. Dozens of engineers and robot enthusiasts worked relentlessly to make it happen. The Carnegie Mellon team was a favorite, but every team faced hurdles—from smashed sensors and exploding toilets to poorly placed tumbleweeds. Was the Grand Challenge too grand for its time? 

The Facts Mark Phillips wrote and performed our theme song. Build Buildings wrote and performed our special ad music. Additional music by Hot Moms Dot Gov, Dream Junkies, 

Bobby Lord, and Haley Shaw. Andrew Dunn mixed this episode. 

Where to Listen


2004 GRAND CHALLENGE RACE MC: All right, Ladies and gentlemen, boys and girls.

LISA CHOW: In 2004, there was a race from California to Nevada. People crowded around the starting line, to watch something that had never been done before.

MC: And off we go...

LISA: All sorts of vehicles—big trucks, small homemade cars, SUVs, and one motorcycle, were competing in a race covering 150 miles. DARPA, the Pentagon’s research agency, was offering $1 million to whoever finished first. And the thing that was so special about these vehicles is that every one of them was driving itself. There was no person at the wheel. No one steering them remotely. These vehicles were making decisions—when to turn, when to slow down—all by themselves.

MC: Making an adjustment, no problem, seeing the k-barrier, adjustment to the left, thank you.

LISA: The race was called the DARPA Grand Challenge, and it’s where some of the biggest names in today's driverless car industry got their start. People with vehicles in the race went on lead projects at Google, Uber, GM, Ford, pioneering a movement that could radically transform our economy and the world.

NEWS ANNOUNCER 1: Major automakers are now busier than ever building vehicles that don’t need you.

NEWS ANNOUNCER 2: ready or not, they're coming. automakers from Stuttgart to silicon valley are in a race to get driverless cars on the roads.

NEWS ANNOUNCER 3: we will almost surely experience a radical shift in how we move, where we live and what our cities look like.

LISA: Billions of dollars are pouring into driverless car research, and the Grand Challenge, this do-it-yourself meetup in the desert, that brought together robot-obsessed hobbyists and inventors, engineers, and high school students—it’s the thing that sparked it all. It was the seminal moment for the industry. But at the time, it was viewed as a total failure.  From Gimlet Media, I'm Lisa Chow, and this is StartUp. Over the course of the next two episodes, we're going to tell the story of the Grand Challenge—the people who shaped it and were shaped by it, and we’ll find out how far we've come in the world of self driving cars.

LISA: One of the best known names from the 2004 Grand Challenge is William Whittaker who goes by Red. Red led one of the teams with vehicles in the race. He’s a professor at Carnegie Mellon. And if you talk to people who know Red, they’ll tell you he’s a legend on campus.  He's a former marine who built robots to clean up the nuclear disaster at Three Mile Island. He races concrete canoes. He holds himself and everyone around him to incredibly high standards. He especially hates when people are late. When I met Red in person, I made sure to be on time.

RED WHITTAKER: Good to see you.

LISA: Good to see you.

LISA: Red is in his late 60s. He’s tall—about six-foot-three. He leads me into his office, and we discuss where to sit.

LISA: Wherever you’re comfortable.

RED: I don’t care about comfort, for god’s sakes, I’m comfortable standing on my head.

LISA: There’s an intensity about Red that doesn't quite come through in his voice. He rarely looks away from you when he talks. At one point he balances his coffee on his knee without ever looking at it. A former student told me that Red challenges his students to take on big, risky, meaningful projects. And he’d do that by telling them about how people in the Arctic would hunt walrus. Red would say to his class—are you going to be a person who goes after the walrus to feed everyone, or are you going to catch a few fish to feed yourself? DARPA’s Grand Challenge became Red’s walrus. Many of Red’s colleagues at Carnegie Mellon believed the Grand Challenge was asking people to do the impossible. The race was going to be in the Mojave Desert, and no one had built a self driving car to navigate that kind of terrain at high speeds. Honestly, the idea sounded crazy. But the Grand Challenge immediately called to Red.

RED:I get that these things are not for everyone. But for gosh sakes you know at least a couple of times in life, let's knock it out of the ballpark and do something for the world.

LISA: Red had a year. He needed to buy a vehicle, and a bunch of expensive computers and sensors. He pulled together enough money to get started. One of his backers was Google, which would later become the first company to seriously invest in self driving cars. Red also had to build a team. The fastest way of doing that was to create a class. He put up flyers all over the Carnegie Mellon campus. They said something like: robot race, million dollar prize, take this class. And dozens of people showed up. One of his key recruits was a 20-year-old from Brooklyn named Matt Johnson-Roberson.

MATT JOHNSON-ROBERSON: I was a junior when I joined. And very quickly it took over my entire academic life and then shortly thereafter my entire life altogether.

LISA: Matt was majoring in computer science at Carnegie Mellon. But so far, everything he'd been studying felt abstract. This felt different. Cars driving themselves, from one place to another, that was something he could talk to his parents about. It felt exciting. But then Matt saw the car they’d be using to build their Grand Challenge vehicle.

MATT: If you can imagine whatever you think in your mind of what a robot should look like you know shiny and new and sort of futuristic if you can imagine the exact opposite of that. They brought in this hummer from the 80s, that was like belching diesel fumes, was loud. It just looked old. And so I was immediately like, oh no this is going to be really really hard.

LISA: The car was a 1986 Humvee that Red bought from a farmer. Red told me the first time he saw it, he knew it was the one. Yes, it was old and beaten, but he says all that mattered was its four wheel drive, the fact that it was high off the ground and could navigate difficult terrain. Red tapped Matt to lead the computer hardware team. The computers would be the vehicle’s control system, and Matt’s job was to figure out not only how to hook them up to the car, so they would run the brakes, gas and steering wheel, but also how to keep them working, even when the car was in motion.

MATT: We were getting server grade hardware. You know, the stuff that should sit in a server room on a rack mount, and figuring out how to stick it on the back of a car, air condition it, and keep it running, particularly when this thing was bouncing around on these dirt tracks

LISA: And then, there were the car’s sensors—which would let it see the world around it. They’d be key to the vehicle’s success, but all the options they found had real problems. Cameras, for example, would give a lot of information about the environment, but the images are flat, making it hard for the vehicle to predict how far away the objects were. LIDAR, another sensing technology, would convey depth and distance, but it wouldn’t give the same rich representation of the world. Matt had been taking other classes but quickly realized he'd have to dedicate all his time to the Grand Challenge so that their car could be ready by race time. So he put everything else on hold, and pretty much stopped sleeping and socializing.

LISA: What did your friends think about what you were doing at this time?

MATT: At first they were kind of, “That’s weird, what are you doing?” And then at some point, they were definitely really worried about me. There is one time where I was walking back to the project area, and I was walking down this really steep flight of step.s I think I fell asleep on my feet, like I was like walking, and then like my brain was like, “Yep, we don't need to be conscious anymore.” And like turned off, and I basically just like almost died falling down this very steep flight of stairs. And I came back and they were like, “What happened to you?”

LISA: All the scrapes and bruises and sleepless nights could be blamed, in part, on the scope of the Grand Challenge. And the fact that DARPA, the race’s organizer, wasn’t sharing much information about what the teams should expect.

MATT: It was so confusing. We were so uncertain about what it was actually going to look like in terms of a challenge. Nobody knew whether, you know, they were going to give us a very easy route or a really hard route. And, you know, it was even more exciting to some degree because of the unknowns

MATT: All that mystery was because DARPA was still working out some of the key details. Tony Tether was the head of DARPA at the time.

TONY TETHER: We really didn't know what we were doing.

LISA; DARPA, best known for being the creative force behind the internet, has a budget in the billions. Its whole mission is to invest in breakthrough technologies for the military. Tony said he created the Grand Challenge because at the time Congress wanted more military vehicles to be unmanned to reduce the number of soldiers killed during basic missions.  But right from the start, pretty much nothing about the race went according to plan. Tony initially wanted the Grand Challenge to go from Disneyland to Las Vegas. But you couldn’t hold a race like this with regular cars and drivers around, so it meant closing down roads and highways. Plus, DARPA would have to pay back business owners, like casinos, for all the money they’d lose while the race was going on. To get an estimate, Jose Negron, the project manager for the Grand Challenge, videotaped intersections on the strip to see how many cars went by every minute and then reported back to Tony Tether.

JOSE NEGRON: That was the fun part is showing him the film, and I said, “Tony just to close down this intersection is going to cost you X amount. This intersection is going to cost you X amount. This intersection is going to cost you X amount.” He goes, “Well, you told me I had 15 intersections. How much more is it going to cost me?” I said, “Well, that's my point.”

LISA: So Tony and Jose decided to move the race outside of big cities. Starting in Barstow, California, and going to Primm, Nevada. But this created other problems. Jose met with a power company that was worried that an autonomous vehicle would veer off course and knock out power lines near Hoover Dam, killing the electricity in Los Angeles. Native American tribes living in the Mojave Desert, they were concerned about unmanned vehicles careening through burial grounds. And Tony says there was another opponent to the race. The U.S. government itself, specifically the U.S. Fish and Wildlife Service.

TONY: They heard what we were doing, and then the next thing we know we have four of them showing up and saying hey what are you guys doing. You know, what you know about the turtles and we all kind of what turtles.

LISA: Tony says turtles but he means tortoises, specifically a protected species called the desert tortoise. To make sure these autonomous vehicles didn’t disturb them, or worse, kill them, Tony agreed that the morning of the race, they’d move all the tortoises off the route.

TONY: I’ll tell you a little thing about desert turtles. It’s dry out there right?

LISA: Uh huh.

TONY: So you say how on earth do they get water? Well they have an ability which is that they regenerate their water. Water goes into them, it turns to urine, they take that urine and actually create it back into water, pure water, internal.

LISA: Wow.

TONY: That’s how they live. And if you pick them up, what happens is that they get frightened, and they pee themselves. I mean they actually pee their urine.

LISA: Oh no.

TONY: And then if you put them down they die, because they’ve lost all the water they were using to recycle. So you could not just go and pick them up. You had to get people who knew how to pick the turtles up. Well I guess I hired all of the biologists in the southwestern part of the United States. Who were who were licensed to pick up turtles.

LISA: While Tony and Jose were dealing with reptiles, government officials, and utility executives, back at Carnegie Mellon, Red Whittaker's team, which had assumed the name the Red team after its leader, was starting to see their 1986 Humvee look more like a robot. They’d named it Sandstorm and it had cyclops looking eyeball on top of it, to help it see the world around it. The team had settled on LIDAR for the car’s main sensor, the same technology that’s big in the self-driving car industry today. And inside Sandstorm sat a large electronics box on shocks, holding the servers, AC units, network switches, and power cables. That stuff, alone, weighed more than a thousand pounds. So the Red Team had built the vehicle. They had all the hardware in place. Now it was time to test the software. To refine the way Sandstorm was interpreting information and maneuvering itself through the world. And to do that they had to move out to the desert, to a setting like the one they’d be racing in. Six weeks before race day, a small group from the Red team moved to a testing facility in Nevada. Kevin Peterson was one of those people. He was from Princeton, New Jersey. A brilliant coder known for his competitive streak.

KEVIN PETERSON: I really remember being in the desert for the first time. And that was it for me. I really fell in love with autonomous vehicles driving outside. I think it's, it's kind of a unique thing. You write some software and you load it on the vehicle and the vehicle comes alive. You can, you can immediately see the effect of what you're doing. I still love that. It's still my passion.

LISA: It was in the desert that the magnitude of what they were trying to do, came into focus. There are so many things that humans do easily that are hard for computers. For example, if I see a puddle in the street, or a plastic bag blowing in the wind, I know that those things are pretty harmless to drive through. But for computers, puddles often look like holes in the street. Floating plastic bags look like flying rocks. It's hard for computers to recognize what kind of object they're looking at, and how exactly to deal with it. In 2004, one of the fundamental challenges for Sandstorm was figuring out: where’s the road. Kevin and the small crew rented RVs, so they could be closer to the testing facility, which was about an hour drive from the nearest town. Matt, the eager junior, had desperately wanted to be part of this group testing Sandstorm in Nevada. He’d already invested so much time in the project and who wouldn’t want to chase a robot car through the desert. So here he was, living his dream, until he saw where he’d be sleeping.

MATT: There were not enough beds in the RV for for even all the people. I was the youngest guy and so I guess I got the short straw. And so I remember I was sleeping on a table in one of the RVs.

LISA: Really?

MATT: Yeah, I think my legs were off it, maybe my head. It was, it was not comfortable. And the first week that we got there we had all this diesel fuel for the Humvee. And we were storing the diesel fuel in the RV and the diesel fuel spilled. So I was sleeping in this RV camper on a table with just diesel fumes like in my eyes and nose. I remember thinking, wow you know I really hope we win.

LISA: They parked the RVs near the two-story garage that housed Sandstorm, where Matt and Kevin would spend most of their days coding. This was early February. The nights were extremely cold. The crew would throw meat on a grill. They subsisted on a lot of bratwurst and peanut butter, which was rough for Matt, who was a vegetarian. And that wasn’t the worst of it. The RVs were equipped with showers and toilets, and at some point they needed to dispose of all the waste. Here's Kevin, the software guy.

KEVIN: They drove into town and all of the water containers were full, and the toilets and the showers just exploded because we drove over rough roads. It was just like wastewater everywhere in these RVs. We had to clean up. It was disgusting. After that we decided that we weren't going to use the bathrooms or the showers in the RV and what that meant was that we really didn't shower for like two months. It was kind of barbaric, really.

LISA: The days followed the same rhythm. Kevin and Matt would code, then take Sandstorm out to test that code. They'd drive behind Sandstorm, watching it turn when the road curved but also watching it crash into fences, rip off wheels, hit large boulders, get stuck in mud. While Matt, Kevin and a handful of others were working hard in the desert, Red Whittaker was desperately trying to raise more money to keep the team alive. He was also leading a group of 30-some people working on Sandstorm's planning, mapping, and perception capabilities back at Carnegie Mellon. But Red took occasional trips out to Nevada to check in on the guys testing Sandstorm.

LISA: Can you can you describe for me seeing them in the desert like what did they look like.

RED: Well I can certainly describe what they smelled like in the desert. There were a lot of hours under a lot of tough circumstances and they were on the grubby side.

LISA: The team did get one break from the desert. The company, Intel, which had donated computers, asked the Red team to present Sandstorm at a conference in San Francisco. Kevin, the software guy, was excited about staying in a hotel, taking a shower. But there was one big problem to work out in this formal unveiling of Sandstorm, because of how the CEO of Intel wanted it to happen.

KEVIN: He wanted sandstorm to drive onto the stage while he was talking about it. Sort of like set the stage. This is the future.

LISA: But that wasn’t going to be possible.

KEVIN: This was in an auditorium and there are crazy radio signals in these auditoriums. And of course at this point the robots didn't, they weren't perfect and they didn't honestly work inside. We ended up putting a person in the car, and they lay down in the car and pushed the gas pedal and the brake with their hand and they couldn't see anything.

LISA: This great triumph of autonomous vehicle technology was being driven by a guy slouched down on the floor.

KEVIN: Craig Barrett, who was the CEO, is up there on the stage in front of all these people and he goes, “And here’s sandstorm.” This music started blaring and the engine screamed and then the vehicle rolled out on stage and then the crowd went nuts.

LISA: The crowd was oblivious to the workaround. The CEO went unscathed. And for the Red Team, it brought home this idea that they were working on something important, something exciting.

KEVIN: It's like one of these sports moments that you don't, you don't really get that in technology. You know there's never like a moment where you like write some software and a thousand people cheer for you. I think we all felt like rock stars.

LISA: Just a few hours later, Kevin and the others were back to the desert, the RVs, the brought-wurst. The big day was just three weeks away, when they’d be competing against more than 20 teams. One of the things the Grand Challenge was testing was speed. This was a race after all. But driving autonomously at high speeds in a complicated environment presented all sorts of challenges. The faster a robot travels, the farther out it has to see. It has less time to make decisions and the decisions it makes have much greater consequences. And so the Red team started testing Sandstorm at higher and higher speeds, to make sure its software was working properly. Things were going well. Four days before the opening ceremony of the Grand Challenge, Matt, Kevin, and a guy named Chris, were out testing Sandstorm at a track they’d been using. Matt, the eager junior, was sitting in a car, watching.

MATT: So it's a massive car barreling around the track, and we're doing 20 miles an hour and then we're doing 30 miles an hour.  And I remember I was sitting in the car with Kevin and we always had walkie talkies, and we were all talking to each other and I think Chris is probably in one of the chase vehicles or somewhere else on the track and he was telling us sort of, “OK, let's go to the next highest speed.” So it's going like 40 miles an hour. And I want to say we get to 50. Now I don't know if you have ever seen an off-road car. I think it was 50 miles an hour. But like a car off road at that speed, it is like tearing around the track and it looks like a music video or like a highlight reel. It's crazy. And me and Kevin are in the car, just incredibly impressed that this is happening and it's amazing. And we're super excited. And I think at that point we’d convinced ourselves that this is going to work. And then Chris is like, “Let's go faster.” And I remember I looked at Kevin and I was like “I don't, this seems crazy.” And he's like, “All right, let's try it.” It was faster than we had ever gone before. And we watched the car going around the track, and we watch it come around the turn and get to the end of its loop. And it jams the steering wheel all the way to the right. It was like slow motion because we watched this giant thing roll over. Like the worst accident you've ever seen on a highway, like this giant thing flips over and then spins. So all of the incredibly expensive and delicate electronics and sensors all sat on the roof. So this custom thing that we had built and spent like the whole year engineering immediately crushed, just completely destroyed. The computers are flying out. There's oil leaking everywhere. The wheels are spinning. Me and Kevin just look at each other and are like, whoa.

KEVIN: We we ran over to the vehicle and started to look at it and tried to figure out you know is the thing salvageable. And it became more and more apparent that things were really really damaged. It's one of those moments where you feel your heart drop.

SPENCER SPIKER: The look in everybody else’s face was defeat. They thought game over.

LISA: That’s Spencer Spiker, a mechanical engineer who was out in the desert with Matt and Kevin. He didn’t see the crash, but he remembers finding out about it.

SPENCER: I don't know how many other folks play sports, but when I played football and you're in the huddle, when you're down you know 15-20 points you can look at your quarterback's eyes and see if he's defeated or not, and then you know there's no winning. But you can tell if there's like the fire there, you know you’ve still got a chance. But in that case you'd look at people and they were just, they were just done

LISA: Or were they. That’s after the break.


LISA: Welcome back to StartUp. So Sandstorm had crashed. Many on the team thought there was no way they'd come back from it. They knew they had to tell Red, who'd dreamed up this whole idea of building an autonomous vehicle in the first place. And so they called to give him the news.

RED: I was here in Pittsburgh. I got enough communication to understand what had occurred.

LISA: What’s your initial reaction hearing this?

RED: It’s a gut-shot. Then, as a leader, it really matters to be clear immediately. What are we going to do. It's not like you're going to show your emotions. I had a sense that somehow Sandstorm’s greatness wouldn't have the time to shine. It was a helluva machine with a lot of capability. And from the moment of the rollover, you know, my sense was, well we’re wounded

LISA: Red hopped on a plane to Nevada. When he got to the testing site, things did not look good.

RED: People were exhausted. People were beaten. There were people splayed out on their backs, completely dead asleep, dead to the world. And it just doesn't work. Like, get up. It's time to go.

LISA: Red had to convince these guys, who were sleep deprived and feeling totally defeated, to give it one more shot.

RED: What I gave was the clear, authentic, unambiguous declaration: Sandstorm’s going to run, it’s going to run in the front. It's going to go the distance. And to get there we've got to do this by sun up and that by tomorrow night and that by this next sundown and that by the next night, and let's haul.

MATT: in the same way that he convinced us that this whole thing was possible...

LISA: Here’s Matt, the junior.

MATT: …he convinced us that a week before the race, that we could rebuild and replicate all of the things that we had spent six months to a year building in a week, and that we could get it back to the point that we were ready to race. And I remember being like, yeah, that makes sense, ok, yeah, of course.

LISA: Red had convinced the team to take on the walrus. In the three days before the Grand Challenge’s opening ceremony, the team worked around the clock. They replaced the hood, rebuilt the engine, got a new sensor, and rebuilt the piece of hardware that was stabilizing the sensor—work that originally took months.

MATT: It was a real testament to the fact that the limits that you think you have to yourself I think aren't really there. Because I thought I was at the exact physical limit of what I was capable of doing. And somehow I think that period we slept even less and worked even harder.

LISA: They finished just in time.  The night before the race, Red wrote in his log, “There is no more practice. Just impeccable execution. Saturday will be a lot of dirt, speed and brutality. We can win this. Spare nothing. Victory or demise.”  At 6:30 on the morning of March 13, 2004, all the teams gathered at the opening gate.

2004 GRAND CHALLENGE RACE MC: And we’re 30 seconds from history.

LISA: The vehicles had staggered start times, so they wouldn’t drive out in a pack and wind up running into each other. First up was the Red team’s vehicle, Sandstorm. The Humvee waited at the starting line, ready to go.

MC: Alright, ladies and gentlemen, boys and girls. OK, the bot has been ordered to run, the green flag waves, the strobe light is on, the command from the tower is to move. Ladies and gentlemen, Sandstorm.

Jose: Sandstorm takes off and it just took off like a bat. Boom.

LISA: That’s Jose Negron, project manager for the DARPA Grand Challenge. Matt, the junior, was with members of the Red Team.

MATT:  I remember seeing Sandstorm leave the opening gates and drive off, and very quickly it was like out of our sight. And I don't think we had ever been in that position before because we had never been at a point where we were not following the vehicle, right. So we had always seen it. And so it was a really weird feeling, and so I remember that very vividly thinking like, oh yeah like it's a self-driving car. It doesn't need us.

TONY: Boy that day, the air was just electric.

LISA: Tony Tether, the head of DARPA, was watching Carnegie Mellon’s vehicle from the stands with a four-star general.

TONY: Carnegie's vehicle takes off, comes down, makes a left-hand turn, takes off, and it's riding down there, and it makes another turn. And he turns to me, and he says, “OK, there's nobody in there?” I said, “Nope, there's nobody inside that.” HE says, “There's nobody like over here someplace remotely controlling?” I said, “No, General, it's doing it all on its own.” And he said, “Wow.” And I thought to myself, “Holy cow.”

LISA: You're impressing a four-star general.

TONY: Well, no, no I was impressing myself. I thought, holy cow. Look at this. When that thing took off, it really, really looked like there had to be somebody driving it, because it was driving itself just like a human was inside. I mean, it was spectacular.

JOSE: When you see an autonomous vehicle coming down toward you charging at 20-30 miles, your head starts thinking, is he going to slow down and turn? Oh my goodness, he’s slowing down and turning, he made it. does he turn again? Yes he does. Does he go across the cattle pins? Yes he did. When you see, that’s a realization that we’re here. We are here.

LISA: While Sandstorm was driving through the desert, other vehicles with names like Avidor, Ghostrider, and The NaviGator took off from the starting line, one by one.

MC: Seven of the vehicles are custom made, three are SUVs, three are six-wheel, and we have a smattering of them that are modified ATVs, and we have one motorcycle.

LISA: The motorcycle was the smallest vehicle in the race. The biggest vehicle of the race...

MC: And off we go with Terramax. Ladies and gentlemen.

LISA: ...was called Terramax.

MC: 32,000 pounds, six wheel military vehicle.

TONY: It took off really good, very impressive.

LISA: That’s Tony Tether.

TONY: Big truck rumbling down, driving this thing by itself. And it got caught. It stopped. It came up against a tumbleweed that had gone into the path. It saw this tumbleweed and it thought, it’s an obstacle. What do I do now? So it starts backing up all by itself. Remember the thing is doing it all by itself. Well in the meantime, a tumbleweed had come behind it, and it had sensors in the back to see where it was backing up, so it sees this tumbleweed, and it says my god, another obstacle, and so it just ends up going back and forth, back and forth. 16 tons stymied by two tumbleweeds.

LISA: If vehicles got far enough out of sight, there was really no way to watch them. The only people who got to see what was happening were the people with DARPA, riding along in chase vehicles, or flying in helicopters overhead, surveying the route. Sandstorm drove one mile, then two, then three, things were looking good. It crossed a highway, got to a cattle gate, drove toward the desert, and then it got to one of the most challenging parts of the course—the switchbacks, incredibly windy roads up and down mountains.

MC: Okay, we just got word that Sandstorm is in the switchback section of the event, here at the DARPA Grand Challenge.

LISA: Matt, the junior, was hanging out by the opening gate area.

MATT:  I remember just not wanting to even hear any updates ‘cause I was so scared that something bad was going to happen.


LISA: This is tape from the DARPA vehicle that was following Sandstorm. They were relaying updates back to Tony Tether.

CHASE VEHICLE: it hit a huge boulder, and it uprooted it and it basically, it almost high centered itself on it, and then it had to back up and get back on course.

LISA: Sandstorm also ran into a few fence posts, but it was at mile seven where the trouble really started.

CHASE VEHICLE: Mic 22, victor 22, robot 22 is smoking. Roger.

LISA: Sandstorm was going up a hill when it drove off the road, and one of its massive wheels slipped into a patch of rocky dirt, and couldn’t get traction.

MATT: One of the things that we had programmed into it was sort of the same level of determination, perhaps, that we had. Which was to basically keep going in any scenario in which it got stuck.

LISA: One of the guys in the chase vehicle got out to take a look at Sandstorm.

CHASE VEHICLE: Mike, be careful cause it might back up.

MATT: It got stuck, and then it was spinning its wheels, essentially, when it went off the road. And it spun its wheels so much that it ripped the tires off the rims and was throwing molten rubber. I guess just onto the track.

CHASE VEHICLE: The left front tire was spinning, spun itself till it ran all the rubber off the tire, and now it’s spinning full bore with no rubber on the tire, there’s no more smoke. You want me to pause it?

PHONE IN CHASE VEHICLE: Victor 21 mic 21. We do want you to pause right now.

CHASE VEHICLE: Ok Tony says disable. Going to go to pause. Mic 22 victor 22 is going to pause, now disable. Alright, we just disabled him. Mic 22 victor 22 is disabled.

LISA: Sandstorm was done. It was stalled out on the side of a ridge, tire spun off, engine smoking. It had gone 7.4 miles, a long way from the one-hundred-and-fifty miles it would’ve had to travel to complete the race. Here’s Matt, again.

MATT: I remember losing it and it being really emotional because it just...it's such a big physical manifestation of your work. It would be like having a fire at an artist’s studio and losing all of your work. Like you know that, yeah, I made that thing, and I could make another thing. But I could never make that exact thing again. And I certainly couldn't replicate whatever the experience was of being part of building it.

LISA: Red Whittaker, he had a different reaction.

RED: When I first heard the news, I had a moment of disappointment, but mostly the idea that you know that's it. That's how far for today.

LISA: While members of the Red Team were working through their let down, there were news reporters waiting … who had no idea the race was even over. They were gathered at the finish line in Primm, Nevada, waiting for the first vehicle to come into view. Tony Tether took a helicopter to break the news.

TONY: I landed there into Primm. And the public affairs person, Jan Walker, came up to me and says, “What are you going to tell them?” And I said, “I don't know.” I guess I got a minute or two to figure that out. So we went in just the two of us. And these cameras, and they say well, “OK, how's it going?” And I said, “It's over.” And they said, “What do you mean it’s over?” I said, “It’s over. The last car caught on fire. And got seven point something miles and that's it. They’re all down.”

LISA: There was no winner that day, but the vehicle that had gone the farthest—it was Sandstorm. Tony said there was silence in the room. The disappointment was clear. So he had an idea.

TONY: I said, “Well, ok. I’ll tell you what though. We’re going to do this again. And we’re going to do it again in about 18 months. And then the prize is not going to be a million dollars. I’m making it $2 million.” Now I did not have the authority to make it that but I said it anyways.

LISA: Luckily, Tony did get what he wanted—another race, another prize, doubling the stakes.

TONY: I did not want it to end, even though I knew that the other guys at DARPA understood this as a Kitty Hawk event. If you go to Kitty Hawk and you see what the Wright Brothers did, it wasn't very spectacular. Their airplane was catapulted. The first flight was 30 seconds. 30 seconds. But in that 30 seconds, The Wright Brothers proved what people really believed at that time couldn't be done. And I always felt that, that Carnegie in that 7.6 miles or whatever the distance was, proved that it could be done. I mean, no other car in history had ever gone anywhere near that. Because people still did not really believe that you can put sensors and computers and into a vehicle and have it make all those decisions you know all by itself.

LISA: The Grand Challenge didn’t only make people believe in the potential of driverless cars, it also shaped the futures of a lot people, including people on the Red team. Matt graduated from Carnegie Mellon the following year and is now a professor at University of Michigan, where he co-directs the school’s center for autonomous vehicles. Kevin, our brilliant coder, he stuck around Carnegie Mellon to pursue his PhD, and is now cofounder of an autonomous vehicle company called Marble. Kevin and Red Whittaker kept at the DARPA races. The next year, the Red Team actually finished the race, but lost to a group from Stanford. The team finally won first place in 2007. They took home the $2 million prize in the third and final race that DARPA organized.

KEVIN: You really had a lot of the capability in 2003, but none of it was together in one place. We were really the first community to work on self-driving cars in a big way, and we were really the first people to crack the challenge. How do you test a vehicle, how do you write the first software, how do you do that safely, how do you develop that software safely. And in 2007 we had the vehicles working, and after that, there was this belief that self-driving cars could come together. Without those races, I really don't think that we would have autonomous vehicles on the road today. I think it would be an entirely different world.

LISA: DARPA’s Grand Challenge inspired one company, in particular. Google. The tech giant hired engineers from the races to start its self driving car project. And for several years after the Grand Challenges, Google was only big company working on autonomous vehicles. They had no idea how it was going to make money. In fact, they called it a moonshot. But then something happened that turned that moonshot into a gold rush.

TRAVIS KALANICK: My name is Travis Kalanick, cofounder CEO of Uber.

LISA: How a race that started in the desert became a much larger race between tech giants and car companies, with billions of dollars in the balance. That's next week on startup.

LISA: StartUp is hosted by me, Lisa Chow. This episode was produced by Bruce Wallace, Luke Malone, Simone Polanen, Emanuele Berry, Amy Standen, and Max Gibson. Our senior producer is Molly Messick. We are edited by Annie-Rose Strasser. And we want to take a moment to thank Emanuele Berry, who is leaving our show to join The Nod, another Gimlet podcast. Our loss is their gain. We’ll miss you, Emanuele. Even though you’re only sitting downstairs from us. Mark Phillips wrote and performed our theme song. Build Buildings wrote and performed our special ad music. For full music credits, visit our website. Andrew Dunn mixed the episode. Special thanks to Michele Gittleman and Vanessa Jameson. To subscribe to StartUp, go to Apple Podcasts, or whichever app you like to use. Or check out the Gimlet Media website: GimletMedia.com. You can follow us on Twitter @podcaststartup. And if you’re curious about what happened to those delicate desert tortoises, they survived the race. Tony Tether made sure all their needs were met.

TONY: In fact we found two turtles who were mating, they were actually on the track. So we didn’t know what the hell to do. We couldn’t go pick them up and move them. And when turtles mate, by the way, they take a long time, so we had to watch them. Because if a car made it as far to these turtles, we were going to have to stop the race, and we thought the cars wouldn’t get there for a couple of hours, and so they must surely be done by then.

LISA: Thanks for listening. We’ll see you next week.

We do our best to make sure these transcripts are accurate. If you would like to quote from an episode of StartUp, please check the transcript with the corresponding audio.