The Backseat Question: Your First Story of a Self-Driving Car

The first time I rode in a fully driverless car, it was 2 p.m. on a overcast October day in 2020, and I was standing in the pickup zone of Phoenix Sky Harbor International Airport, dragging a dented 24-inch suitcase behind me and second-guessing every life choice that had led me to tap “request ride” on the Waymo app instead of calling a regular Uber. I’d watched dozens of TikTok videos of people hailing autonomous taxis in Phoenix, but watching strangers do it on my phone was very different from standing in front of a white Jaguar I-PACE with no driver behind the wheel, no steering wheel visible through the front windshield, and a small Waymo logo glowing on the side door. I’d half expected a guy in a Waymo hoodie to pop out of the backseat and say “Gotcha!”, but when I pulled the door handle, the backseat was empty, save for a touchscreen mounted on the front console displaying a map of my route to a hotel in downtown Scottsdale. The door locked shut with a soft click before I could even sit down, and a robotic voice chirped from the screen: “Please fasten your seatbelt to begin your trip.” My palms were sweating so bad I almost slipped on the leather seat.

The car didn’t lurch or honk or do any of the chaotic things I’d seen new drivers do when I took my driver’s ed class at 16. It just eased forward, slow and steady, like it was being pulled by an invisible hand that had decades of driving experience. I gripped the back of the front seat so hard my knuckles turned white as we merged onto the airport highway, and my brain was racing: *How does it know where the lane lines are? How does it know not to hit that delivery truck pulling out of the parking lot? Does it even know I’m back here?* Then, half a mile down the road, the car flicked its left turn signal and drifted smoothly into the left lane, no sudden swerve, no blaring horn from the truck we were passing. I blinked. The right lane ahead was blocked by a row of orange construction cones and a guy in a neon vest holding a stop sign. The car had seen that before I had.

Five minutes later, a delivery guy on a moped cut across three lanes of traffic to make a right turn, and I yelped so loud the car’s interior lights flickered. I braced for a crash, but the car didn’t even honk. It just slowed down, gentle as a grandfather braking for a cat in the road, and came to a stop 6 feet behind the moped. The delivery guy waved an apology, gunned his moped down the side street, and the car accelerated back up to highway speed, smooth as if nothing had happened. I stared at the front screen, where a 3D rendering of the road glowed in blue and green: little red dots for other cars, a tiny orange dot for the moped, bright green lines for the lane markers. It had seen the moped before I did. It had decided to stop before I could even process what was happening. It had done all of this without a brain, without hands, without a single conscious thought.

That 15-minute ride from the airport to my hotel rewired how I thought about cars. For 20 years, I’d driven everywhere, and I’d always thought of driving as a deeply human skill: you look at the road, you make judgment calls, you react to other people’s mistakes, you know when to speed up and when to slow down based on a gut feeling you can’t even explain. But that ride made me realize that all of that “gut feeling” isn’t magic. It’s just a set of steps your brain runs automatically, without you even noticing, every time you get behind the wheel. And if you can break those steps down into small, repeatable parts, you can teach a computer to do them too.

That’s exactly what self-driving car engineers have done. The “magic” of a car driving itself isn’t magic at all: it’s four separate, interconnected technology modules working together in perfect sync, 100 times a second, every second the car is moving. Together, they do exactly what a human driver does: they see the world around them, decide what to do, plan how to do it, and move the car to make it happen. The only difference is that these modules never get tired, never get distracted by a text message, never get road rage, and never have to pee in the middle of a long trip.

For the rest of this book, we’re going to follow the exact same journey you took in that Waymo from the Phoenix airport. We’ll break down every single thing the car did on that ride—every merge, every stop, every lane change—into the work of those four modules, and we’ll explain how each one works, no engineering degree required. You don’t need to know how to code, you don’t need to know calculus, you don’t even need to know the difference between a CPU and a GPU. All you need is curiosity, and the same willingness to learn that made you look up from your phone on that first ride to wonder, *How the hell does that thing work?*

Before we dive into the nitty gritty of how cars see, think, plan, and move, let’s get one thing straight right off the bat: self-driving cars are not perfect, and they’re not the all-powerful, all-weather machines you might have seen in sci-fi movies. To make sure we’re all on the same page, let’s bust a few of the most common myths that get thrown around every time someone mentions autonomous vehicles.

> 🚗 Myth-Busting Lane: 5 Things You Probably Got Wrong About Self-Driving Cars > > Headlines love to overhype or fearmonger when it comes to self-driving tech, so let’s cut through the noise with cold, hard facts: > > 1. Myth: Self-driving cars can drive anywhere, in snow, rain, or shine. > *Reality:* The vast majority of commercial self-driving taxis operating today are limited to pre-mapped, urban neighborhoods in sunny, mild-weather cities: Phoenix, San Francisco, Los Angeles, Austin, and Beijing are the biggest hubs right now. Heavy rain, snow, fog, or unmapped backroads will shut most of them down instantly: rain blocks camera lenses and scatters laser signals from LiDAR, snow covers the lane lines the car uses to navigate, and unmapped roads don’t have the centimeter-accurate data the car needs to know exactly where it is. Think of it like a new driver who’s only ever practiced in their hometown: they’re great on local streets, but drop them in a blizzard in a Rocky Mountain town with no street signs, and they’re completely lost. > > 2. Myth: Self-driving cars never get into accidents. > *Reality:* They’re far safer than human-driven cars in most scenarios, but they’re not crash-proof. A 2023 report from the U.S. National Highway Traffic Safety Administration (NHTSA) found that 94% of all traffic accidents in the U.S. are caused by human error: speeding, distracted driving, drunk driving, fatigue. Self-driving cars eliminate almost all of those risks, because they never get distracted, never speed, never drink, and never get tired. But they can still crash if their sensors miss something, or their algorithms make a wrong call. In 2023, a Waymo taxi in San Francisco was T-boned by a delivery truck that ran a red light, and a Cruise self-driving car in 2022 hit a pedestrian who was walking in front of it at night, because its sensors failed to detect them in time. > > 3. Myth: Self-driving cars don’t need any human oversight. > *Reality:* The fully driverless taxis you can hail today are classified as Level 4 autonomous, meaning they can drive themselves in specific, pre-defined areas without a human driver behind the wheel. But every single one of them has a remote safety team monitoring their drives 24/7, ready to take over remotely if the car runs into a problem it can’t solve. For consumer cars with Level 2 or Level 3 autonomy (like Tesla’s Autopilot or Mercedes’ Drive Pilot), you still have to keep your hands on the wheel and be ready to take over at any time. > > 4. Myth: Self-driving cars are way more dangerous than human-driven cars. > *Reality:* As of 2024, Waymo’s fully driverless taxis have a crash rate 85% lower than human-driven cars in the same areas