Waymo has taught cars to navigate intersections, pedestrians, highways and some of America’s most complicated urban streets without anyone behind the wheel. Yet Waymo parking tickets reveal a strangely ordinary problem that may become increasingly important as robotaxis scale: after solving how to drive through the city, autonomous vehicles still have to figure out exactly where to stop.
That sounds minor until thousands of vehicles start making pickups and drop-offs every day. Waymo has already moved into the phase where lowering vehicle costs matters, as its Ojai robotaxi strategy demonstrates. But cheaper autonomous vehicles are useful only if cities can absorb thousands of them without every restaurant, hotel, bike lane and bus stop becoming a negotiation over curb space.
Waymo Parking Tickets Reveal a Different Scaling Problem
Waymo vehicles received 8,309 parking citations in San Francisco from January 2025 through mid-July 2026, carrying $957,008 in fines.
The San Francisco citation data shows 4,178 tickets during 2025 and another 4,131 through July 19, 2026. By comparison, Waymo received 589 parking citations during all of 2024.
The most common violation was parking in designated stands reserved for particular vehicle types, with 2,038 citations. Other frequent problems included prohibited parking, obstructing traffic, double parking and blocking bicycle lanes.
Those raw numbers require context. Waymo’s service has expanded rapidly, so more citations do not automatically prove each individual robotaxi has become worse at curb behavior.
Without a comparable ticket-per-trip or ticket-per-mile measure, the data cannot establish that.
What it does prove is more important: small operational problems become large at scale.
The Curb Is Harder Than It Looks
Driving rules are often comparatively explicit.
A traffic light turns red. A lane has boundaries. A pedestrian enters a crosswalk. A speed-limit sign displays a number.
The curb is different.
One stretch may permit passenger loading for five minutes. Twenty feet away could be a commercial zone, bus stand, bike lane, driveway, fire hydrant or no-stopping area. Restrictions can change depending on the hour.
San Francisco’s passenger loading rules demonstrate that complexity. White zones are intended for active passenger loading, generally limit vehicles to five minutes and require drivers to remain with the vehicle under normal circumstances.
Then add human behavior.
A rider may not be standing outside when the robotaxi arrives. Someone may request a pickup directly in front of a crowded restaurant. Delivery vehicles could occupy the legal loading area. A cyclist may be approaching exactly when the passenger wants to open a door.
The vehicle does not merely need to understand regulations.
It has to interpret a constantly changing curb.
Human Taxi Logic and Robotaxi Logic Are Not the Same
Humans frequently solve curb problems by improvising, including in ways that are technically illegal.
That creates an awkward benchmark for autonomous vehicles.
| Curb Situation | Human Driver Response | Robotaxi Problem |
|---|---|---|
| Passenger is not ready | Wait briefly or circle | Must choose between waiting and rerouting |
| Legal curb is occupied | Stop nearby or double park | Illegal stop can trigger enforcement |
| Bike lane beside destination | Judge traffic and stop opportunistically | Must protect cyclists while serving rider |
| Busy restaurant entrance | Follow local driving behavior | Informal behavior may conflict with rules |
| Loading restriction changes by time | Read sign and interpret context | Software must understand location and timing |
| Passenger requests exact doorstep | Negotiate a practical compromise | Vehicle must select a safe legal alternative |
This exposes a fundamental difference between autonomous driving and autonomous taxi service.
A robotaxi does not simply need to transport someone safely from coordinate A to coordinate B. It needs to provide a pickup and drop-off experience convenient enough that passengers continue using it.
Make the stop too conservative and riders complain about walking.
Make it too convenient and the vehicle may obstruct everyone else.
Bigger Fleets Turn Edge Cases Into Daily Operations
Waymo’s expansion makes this problem more urgent.
The company began welcoming public riders in Denver, San Diego and Tampa on September 1, bringing fully autonomous rides to 14 cities. Its latest service expansion means the challenge is no longer confined to a handful of carefully controlled neighborhoods.
Every new city adds different curb markings, enforcement cultures, street layouts and local habits.
A strange pickup situation that occurs once every several thousand trips sounds manageable during testing. At commercial scale, rare situations repeat constantly.
This is one of autonomy’s most underestimated mathematical problems.
Scale multiplies the exceptions.
The same applies to construction zones, emergency vehicles and unusual traffic control. But curb management is particularly difficult because every robotaxi ride eventually requires the car to perform the maneuver twice: once when the passenger enters and once when the passenger leaves.
Cities May Need to Become Part of the Robotaxi System
The obvious response is better software, but software alone may not solve the issue.
Cities may eventually need more dedicated passenger-loading zones, clearer digital curb maps and infrastructure capable of telling autonomous fleets which spaces are available in real time.
That would represent a major shift in how robotaxis are discussed.
For years, autonomous-driving companies have treated the city as an environment their technology must learn to understand. Large-scale deployment could reverse part of that relationship, forcing cities to redesign pieces of their infrastructure around autonomous fleets.
There is already precedent in ride-hailing. Airports, hotels and entertainment districts frequently establish dedicated pickup zones for Uber and Lyft because thousands of individual curbside requests become difficult to manage.
Robotaxis could intensify the same pressure.
A city filled with autonomous taxis may eventually need machine-readable curb space as much as better vehicle intelligence.
The Last 20 Feet Could Decide Whether Robotaxis Feel Normal
Waymo’s nearly $1 million in San Francisco fines will not determine the future of autonomous driving. For a company backed by Alphabet, the money itself is hardly the interesting part.
The behavior behind those fines is.
Robotaxis are moving from engineering demonstrations into transportation infrastructure. At that scale, success will be judged not only by whether the vehicle avoids crashes but by whether it can perform thousands of mundane tasks without irritating riders, cyclists, bus operators, pedestrians and everyone stuck behind it.
That is why the Waymo parking tickets story matters.
The autonomous-car industry spent years focused on the spectacular challenge of removing the driver. The next phase may revolve around problems that sound almost embarrassingly ordinary: where to wait, where to pull over and how long to remain there.
Waymo has shown that a computer can drive across San Francisco without a human holding the steering wheel.
The harder commercial test may be teaching that computer exactly which 20 feet of San Francisco it is allowed to occupy when the ride ends.
