What Waymo's Actually Doing Right Now
Waymo operates autonomous ride-hailing vehicles in a handful of U.S. cities, and the service area has been shifting pretty frequently. Right now their main paid operations run in parts of Phoenix, San Francisco, Los Angeles, and they've announced plans for Austin and other markets. The technology itself is a combination of LiDAR arrays, radar, cameras, and a fairly beefy computer stack that runs on proprietary hardware. It's not a single algorithm you can download and run on your laptop. There's no installer, no APK, no software package you grab from a website. It's a full vehicle platform built around a modified Chrysler Pacifica or a custom-bodied electric vehicle called the Waymo Driver. The system works by having each vehicle continuously map its surroundings, predict what other road users will do, and plan a path that avoids conflicts. It's a sensor fusion approach rather than relying on any single data type. The LiDAR gives them precise 3D point clouds of the world around the car. Radar handles velocity estimation on objects that might be hard for cameras to track. Cameras feed visual classification into the stack. All of that data gets processed in real time by GPUs mounted in the vehicle. What most people don't understand is that the vehicle also has a remote operations center. If the car encounters a situation it can't resolve on its own, a human operator can step in through a live video feed and guide it out of the predicament. This happens rarely, maybe a few times per thousand miles, but it's there as a safety net. The company calls this "remote assistance" and it's a key part of why the service can operate without a human driver in the seat.
I used the service regularly in San Francisco for about eight months last year while my car was in the shop. The one edge case that still bothers me happened on a stretch of Mission Street where construction had rerouted traffic into a narrow lane bounded by concrete barriers on one side and a row of parked cars on the other. The Waymo vehicle identified the barriers correctly but treated the parked cars as potential obstacles too aggressively. It slowed to a crawl and then came to a complete stop in the middle of the narrowed lane, blocking traffic behind it. The remote operator eventually stepped in and talked the car through inching forward at two miles per hour while I watched through the app's live camera feed. The workaround was to message the support team through the app before hailing the next ride, flagging the construction zone so they could prep the route proactively. On subsequent attempts, the car handled it fine once the route was adjusted. This is the kind of thing that doesn't make it into press releases. The technology is impressive but it still stumbles on unstructured construction zones, unpredictable human drivers, and certain weather conditions that degrade sensor performance. Rain is actually less of a problem than you'd think because LiDAR penetrates light precipitation fine. Heavy fog and snow are where things get messy. The system will usually refuse to operate or request that you end the ride early if conditions drop below a certain visibility threshold. There's also a misconception about how Waymo defines its service areas. It's not a single continuous zone where you can hop in anywhere. Each city has geofenced operational design domains, meaning the car only runs within a specific boundary. If you try to hail a ride outside that boundary, the app won't let you. The boundaries have been expanding over time but they still cover roughly neighborhood-scale areas rather than entire metropolitan regions. In Phoenix, for example, the service area covers parts of Scottsdale and central Phoenix but doesn't extend to the airport or most suburban areas.
The app itself is straightforward. You create an account, link a payment method, and request a ride the same way you would with Uber or Lyft. The main difference is you don't get to choose a driver or see their rating. You just see the vehicle approaching and get a photo of the actual car with its license plate. When you enter the vehicle there's no one in the driver's seat. There are two cameras inside the cabin and a soft voice prompt that tells you to fasten your seatbelt and that the ride is beginning. If you need to communicate with support there's a button in the app for that. The pricing structure has been shifting as Waymo scales. Early on they offered promotional rates that were significantly cheaper than ride-hailing alternatives. Now in most markets the pricing is closer to standard UberX or Lyft rates, sometimes slightly above depending on demand. There's no surge pricing in the same way, but prices do adjust based on time of day and local demand patterns. A typical trip in San Francisco might run between eight and fifteen dollars for something that would normally cost twelve to twenty on conventional ride-hailing platforms. One thing that genuinely surprised me after using both Waymo and traditional ride-hailing extensively is how much smoother the ride quality tends to be. The autonomous system accelerates and brakes more gradually than most human drivers. It doesn't slam on the brakes when the car in front taps its brakes. It doesn't accelerate hard from stops unless absolutely necessary. For people who get motion sick in cars, this makes a noticeable difference. I've had friends who can't tolerate regular ride-hailing because they get nauseous, and they told me Waymo rides are the first time they've been able to use the service without feeling sick.
Get the Full Details
The downside is availability. Outside the core service areas, you're out of luck. There's no workaround for this other than waiting for expansion. The company has been relatively quiet about timeline specifics for new cities, which makes planning difficult if you're trying to decide whether to rely on it or keep a backup option. I always kept Uber and Lyft installed as fallbacks, and I wish I hadn't needed to but the coverage gaps made it necessary. In San Francisco the coverage was decent during weekday daytime hours but spotty at night and on weekends in certain neighborhoods. If you're outside a service area and want to get a sense of whether expansion is coming to your city, the most reliable source is Waymo's official announcements on their website. There's no third-party tracking tool that's accurate, and various forums have people speculating about dates that turn out to be wrong. The company has historically been cautious about committing to timelines publicly. Another thing worth noting is the safety record. Waymo publishes annual safety reports and the data shows their autonomous vehicles are involved in fewer accidents per mile driven than the average human driver in the same cities. Most of their incidents are minor, low-speed contact events rather than serious collisions. But "fewer accidents" doesn't mean zero accidents, and there have been situations where a Waymo vehicle stopped unexpectedly in a travel lane or hesitated at intersections in ways that confused other drivers. The remote assistance team tracks these events and uses them to improve the system, but the feedback loop isn't instant. You might ride in a car that hasn't been updated with lessons from a similar situation that happened an hour ago in another part of the city.
For people considering making this a regular transportation option rather than a novelty, the practical advice is simple. Check the current service map on Waymo's site before you commit. Test it on a short trip during daylight hours first. Have a backup plan for when the car can't reach your destination. Don't expect it to replace your car entirely unless you live squarely within an active service area and your daily needs are mostly within that zone. And if you do encounter something weird, document it in the app. The support team actually reads those reports and routes them to the right engineering team. It's not a black hole like some companies' support channels.
