Waymo Explained: How the Robotaxi Pioneer Really Works

Waymo is the trailblazer of fully driverless mobility: no safety driver, complex intersections, dense city traffic. Here's how the Waymo Driver actually turns LiDAR, cameras, radar, and massive amounts of data into real-time driving decisions, and what it means for the future of getting around.

Waymo explained: What makes the robotaxi pioneer different?

No name is more synonymous with self-driving technology than Waymo. Born out of a Google research project back in 2009, the company now runs fully driverless robotaxi fleets across several US cities, with no safety driver on board. Anyone following the EV space eventually runs into Waymo, because autonomous driving and electric propulsion are closely linked, both technically and strategically. Nearly every major robotaxi fleet today runs on electric vehicles, since sensors, onboard computing, and power management integrate far more easily into an electric drivetrain than into a combustion engine.

Waymo is worth a closer look for several reasons. First, it shows just how far the technology has actually progressed, not as some future promise, but as a real, paid service carrying passengers every day. Second, running a large autonomous fleet raises questions that touch charging infrastructure directly: who charges thousands of driverless vehicles, and how does autonomous charging work when there is no driver left to plug the car in? These questions will become increasingly relevant outside the US too, even in markets where regulation still lags well behind.

Before diving into the technical details, a quick note on classification: Waymo operates at one of the highest rungs of the SAE levels of autonomous driving, specifically Level 4. That means the system drives entirely on its own within a clearly defined operating area, with zero human intervention, but not everywhere and under all conditions the way a hypothetical Level 5 vehicle would. That distinction explains why Waymo still limits its service to a handful of carefully chosen, painstakingly prepared cities.

Source: The Wall Street Journal: Watch Waymo's Self-Driving Taxis Put Urban Terrain to the Test | WSJ

How the Waymo Driver works: sensors, maps, and decision logic

At the heart of every Waymo vehicle sits the Waymo Driver, a complete hardware-and-software system that handles perception, decision-making, and vehicle control all by itself. This isn't a driver-assist feature that makes highway driving more comfortable, it's a full replacement for the human driver, navigating dense city traffic on its own. A Wall Street Journal ride-along report shows exactly how this system handles real street scenarios in Chandler, Arizona, and San Francisco, with nobody behind the wheel at all.

Three sensor types that cover each other's blind spots

Perceiving the surrounding environment relies on three different sensor technologies working together, each bringing its own strengths to the table. For a deeper technical breakdown of how these systems work and differ, see our guide on LiDAR, radar, and cameras. For Waymo specifically:

• LiDAR uses laser pulses to build a high-resolution 360-degree 3D picture of the surroundings, with a range of up to roughly 300 meters (about 1,000 feet).
• Cameras provide a similarly complete 360-degree view and can pick out traffic lights, road signs, and lane markings from over 500 meters (roughly 1,600 feet) away.
• Radar measures the speed and distance of nearby objects and keeps working reliably even in rain, fog, or snow.

This so-called sensor fusion means the weaknesses of any one system, say, a camera's limited visibility in bad weather, get compensated for by the others. Only by combining laser, image, and radio-frequency data multiple times per second does the vehicle build a robust, redundant picture of everything around it, with each sensor cross-checking and correcting the others in real time.

Ultra-detailed maps as a second layer of perception

Before a Waymo vehicle is ever allowed to drive in a new area, the company maps that zone in extreme detail ahead of time. Lane lines, curbs, traffic light positions, and crosswalks all get captured and stored in a high-precision digital map. During actual driving, the vehicle constantly compares its live sensor data against this pre-built map. That comparison lets the system instantly spot anything that's changed, a fresh construction zone, say, or a car parked somewhere it shouldn't be, and react accordingly. This heavy upfront preparation is exactly why Waymo expands into new cities gradually rather than launching everywhere at once, and why every new city spends months being driven by test vehicles with human safety drivers before driverless operation ever begins.

Sense, Solve, Go: the decision logic behind the wheel

Waymo's actual driving decisions follow a clear framework the company calls "Sense, Solve, Go." The system continuously answers four core questions:

1. Exactly where am I right now?
2. What's in my immediate surroundings?
3. What's likely to happen next?
4. What should I do about it, right now?

These four steps run multiple times per second and form the basis for every single steering, braking, or acceleration decision. Making those calls reliably requires artificial intelligence that's continuously trained and refined. Exactly how AI and machine learning get applied to self-driving cars is a rich topic in its own right.

A massive body of driving experience as training data

Waymo trains its AI models on an enormous dataset: hundreds of millions of real-world miles driven on public roads, plus billions more miles run in simulation. Those simulations matter enormously for training the system on rare but dangerous scenarios that only pop up occasionally on actual streets, a child suddenly darting into the road, or a driver going the wrong way at an intersection. Without combining real-world data with simulation at this scale, it would be nearly impossible to train a system that responds appropriately to genuine edge cases.

Waymo on the road: Chandler, San Francisco, and the safety question

Chandler, Arizona, was one of the first zones anywhere where Waymo operated completely driverless, with no safety driver in the vehicle at all. Only after extensive testing there did the company gradually expand to far more demanding cities like San Francisco, where narrow streets, heavy pedestrian and cyclist traffic, and complicated intersections present entirely different challenges than Arizona's wider, more spread-out roads. The test drives shown in the WSJ report make clear just how different these two environments are for an autonomous system to handle. Waymo now runs its Waymo One ride-hailing service in Los Angeles, Phoenix, San Francisco, Austin, and numerous other US cities, logging around half a million paid rides per week (as of mid-2026, with Waymo aiming to reach a million a week by year's end) and a cumulative total of autonomous miles driven that has now surpassed 200 million.

Despite this progress, interactions with pedestrians and cyclists at intersections remain a key test of how reliable robotaxi systems really are in dense urban traffic. Independent reports and safety data keep this conversation going, and Waymo itself regularly publishes figures on crashes and near-miss incidents. Anyone wanting a deeper look at the current data can find it in our guide on how safe self-driving cars actually are, based on studies and hard numbers. The bottom line: autonomous systems like Waymo get held to an extremely high bar here, because even a single serious incident can shake public trust in the entire technology.

If you're wondering what a robotaxi actually offers over a regular taxi or a personal car day to day, our guide on what a robotaxi is lays out the basics. In short: a robotaxi is a driverless vehicle you hail through an app that takes passengers from A to B with zero human control at the wheel, while the operator manages the fleet, maintenance, and software centrally behind the scenes.

What's next for Waymo, and what does it mean elsewhere?

Waymo keeps expanding into new American metros and has already begun offering higher-speed freeway rides in several cities, though it temporarily paused freeway service in mid-2026 to refine how the system handles tricky conditions like construction zones. At the same time, the company is pouring resources into even more capable sensors and larger simulation environments to push reliability further and eventually operate in regions with harsher weather, like heavy snow or torrential rain. Meanwhile, rivals like Tesla, Zoox, and Baidu's Apollo Go are watching the market closely, and competition in autonomous mobility is only going to intensify over the next few years.

In Germany, the picture looks quite different. German law is considerably stricter than US regulation on this front, and a nationwide driverless robotaxi rollout without a safety driver isn't realistic anytime soon. Our guide on robotaxi laws in Germany and Europe covers the current legislative state, pilot programs, and planned timelines in detail. While American cities already run commercial robotaxi fleets, German projects are mostly still stuck in testing and approval phases, often coordinated closely with regulators such as Germany's Federal Motor Transport Authority (KBA).

Still, EV owners and shoppers everywhere have good reason to pay attention to Waymo, because many of the technologies described here, from sensor fusion to high-precision mapping to AI-driven decision-making, are steadily making their way into the driver-assist systems found in mainstream cars too. Anyone interested in the broader shift toward vehicle automation should also get familiar with the basics of EV charging, since autonomous mobility and electric charging are only going to grow more intertwined, for instance once fleets start scheduling their own charging sessions and pulling up to chargers without any human involved. Waymo already offers a pretty realistic preview of how one piece of that connected mobility future actually works in practice.

Additional Video

For a deeper look behind the scenes, this Bloomberg interview with Waymo's VP of onboard software explains how the company actually engineers and builds its self-driving system, complementing the on-street testing footage from the first video.

Source: Bloomberg Originals – How Waymo Builds Self-Driving Cars