Waymo self-driving cars: the struggle the standoff and what it means for autonomy

01What is happening with Waymo cars

Waymo LLC is an American autonomous driving technology company headquartered in Mountain View, California. It is a subsidiary of Alphabet Inc., Google's parent company. As of June 2026, Waymo operates public commercial robotaxi services in 10 US metropolitan areas, has 3,871 robotaxis in service, provides 500,000 paid rides per week and had logged

Waymo, the autonomous driving subsidiary of Alphabet, operates commercial robotaxi services in Phoenix, San Francisco, Los Angeles, and Austin. The company has driven millions of autonomous miles and carries paying passengers without a human safety driver. However, recent incidents have drawn public attention to scenarios where the vehicles stop unexpectedly, block intersections, or require remote operator intervention.

In Los Angeles, videos have surfaced of Waymo vehicles stopping in the middle of busy intersections, refusing to proceed, and creating traffic disruptions. In some cases, the vehicles appear to be stuck, honking at each other or blocking emergency vehicle access. These incidents, while not resulting in collisions, reveal the limitations of autonomous systems in handling complex, unpredictable urban driving situations.

02The scenarios where self-driving fails

A self-driving car, also known as an autonomous car, driverless car, robotic car, or robo-car, is a car that is capable of operating with reduced or no human input. They are sometimes called robotaxis, though this term refers specifically to self-driving cars operated for a ridesharing company. As of 2026, the term "self-driving" lacks an agreed st

Autonomous vehicles use a combination of sensors, including lidar, cameras, and radar, to perceive their environment, and machine learning models to interpret sensor data and make driving decisions. The system performs well in well-mapped, predictable environments. It struggles with novel or ambiguous situations: construction zones with temporary lane changes, unusual traffic configurations, pedestrians behaving unpredictably, and complex multi-vehicle interactions at intersections.

One common failure mode is overconservative behavior. The vehicle detects a potential conflict, such as a pedestrian near the crosswalk, and stops even when a human driver would proceed safely. In dense urban traffic, this can cause the vehicle to stop in places where stopping creates a hazard, such as the middle of an intersection after the light has changed. The vehicle is behaving safely according to its own risk model, but the result is unsafe in the context of surrounding traffic.

03How cities are responding to autonomous vehicles

Cities where Waymo operates have had varied reactions. San Francisco initially resisted the expansion of robotaxi services, citing safety concerns after Cruise vehicles blocked intersections and interfered with emergency responders. The city eventually allowed Waymo to expand after the company demonstrated improved performance and addressed specific incidents. Los Angeles has taken a more permissive approach but is monitoring the situation as complaints accumulate.

Some municipalities are considering regulations that would allow them to restrict or suspend autonomous vehicle operations in response to repeated disruptions. The tension is between encouraging innovation and protecting public safety and traffic flow. Autonomous vehicle companies argue that their safety record is better than human drivers, while critics point to incidents that human drivers would not cause, such as stopping in intersections.

The infrastructure challenge is also significant. Autonomous vehicles are designed for existing road infrastructure, but their behavior interacts with traffic signals, lane markings, and road geometry in ways that cities may need to adapt. Some cities are exploring dedicated lanes or zones for autonomous vehicles, while others are updating signal timing and signage to accommodate their specific behavior patterns.

04The safety record compared to human drivers

A self-driving car, also known as an autonomous car, driverless car, robotic car, or robo-car, is a car that is capable of operating with reduced or no human input. They are sometimes called robotaxis, though this term refers specifically to self-driving cars operated for a ridesharing company. As o

Waymo has published safety data suggesting that its vehicles have a lower crash rate per million miles than human drivers in the areas where it operates. The company reports that its vehicles are involved in fewer injury-causing crashes and fewer airbag-deploying crashes than the human-driven baseline. However, the comparison is complicated by operational domain restrictions: Waymo operates primarily in mapped areas, during favorable weather, and avoids the most challenging scenarios that human drivers navigate daily.

The safety comparison also depends on what is counted. A Waymo vehicle stopping in an intersection does not register as a crash, but it can cause secondary incidents as other drivers swerve or brake unexpectedly. Near-misses, traffic disruption, and emergency vehicle interference are not always captured in crash statistics but affect public safety. A comprehensive safety assessment needs to account for both crash rates and system reliability in traffic flow.

05Why some situations remain unsolved

The fundamental challenge of autonomous driving is the long tail of rare scenarios. A system that handles 99 percent of driving situations correctly still encounters the remaining 1 percent frequently in aggregate, and those edge cases can be dangerous. Construction zones, emergency vehicles with unusual traffic patterns, police directing traffic, severe weather, and novel road configurations each represent a category that the system may not have encountered in training.

Machine learning systems are statistical pattern matchers. They learn from large datasets of driving scenarios, but they cannot reason about a situation they have never seen. A human driver encountering a novel situation, such as a sinkhole in the road or a parade crossing an intersection, can use general intelligence to decide what to do. An autonomous system that has not been trained on that specific scenario may stop, which is often the safest default but can create traffic problems.

The industry is divided on how to address the long tail. One approach is to collect more data and train on more scenarios, gradually expanding the coverage of the system. Another is to build more general reasoning into the driving system, using larger AI models that can understand context and intent rather than just pattern matching. Both approaches are active areas of research, and neither has definitively solved the problem.

06The regulatory landscape for self-driving

The regulatory framework for autonomous vehicles is fragmented. The National Highway Traffic Safety Administration regulates vehicle safety standards at the federal level, but states regulate vehicle operation, licensing, and insurance. This creates a patchwork where a vehicle approved in one state may not be legal in another. Companies like Waymo must navigate different requirements in each city where they operate.

The California DMV requires companies testing autonomous vehicles to report disengagements, instances where a human safety driver had to take control. These reports provide a window into system performance, though critics note that the metric is self-reported and does not capture all failure modes. The data consistently shows Waymo with the lowest disengagement rate among companies with significant testing mileage, but the rate varies by operating domain and conditions.

Future regulation may address the gaps in current oversight. Proposals include standardized safety metrics for autonomous systems, mandatory reporting of near-misses and traffic disruptions, and federal frameworks that preempt state-by-state fragmentation. The regulatory environment will significantly influence how quickly and where autonomous vehicle services can expand.

07What the future of autonomous transport looks like

The near-term future of autonomous transport is likely to be gradual expansion in favorable operating domains. Waymo and competitors will add cities, expand service areas, and increase fleet sizes, but primarily in places with good weather, well-mapped roads, and manageable traffic complexity. The vision of a fully autonomous vehicle that can drive anywhere a human can is years away.

The business model is also evolving. Robotaxi services charge per ride, competing with human-driven rideshare. The economics depend on the cost of the vehicle, the utilization rate, and the cost of remote support infrastructure. If autonomous rides are cheaper than human-driven rides, the market could grow rapidly. If the cost of maintaining the fleet, remote operators, and mapping infrastructure is too high, growth will be slower.

Beyond passenger transport, autonomous technology is being applied to trucking, delivery, and industrial applications. These domains may prove more tractable than urban robotaxi service because the routes are more predictable and the operating environment more controlled. The future of autonomy may be uneven, with some applications scaling quickly while the general-purpose robotaxi remains a work in progress.