When robot taxis get stuck, a secret army of humans comes to the rescue

In cities where autonomous robotaxi services are now a daily presence on busy streets, there is a hidden workforce quietly ensuring that these driverless cars continue to operate. Despite much of the public conversation about self‑driving technology focusing on artificial intelligence and automation, the reality on the ground reveals that humans still play an essential role behind the scenes, stepping in whenever robotaxis encounter simple issues they cannot resolve on their own.

Waymo, a major self‑driving ride‑hailing service, operates hundreds of robotaxis in cities such as Los Angeles and San Francisco. These vehicles are designed to transport passengers without a human driver by using advanced sensors, artificial intelligence and preloaded maps to navigate complex urban environments. In most situations, the vehicles perform as advertised, transporting riders with efficiency that rivals or even exceeds traditional taxi services. However, when a minor problem such as an improperly closed door or a drained battery arises, the robotaxis can become stranded, helpless to fix the issue on their own.

These limitations are not rare technical flaws so much as inherent constraints of current autonomous systems. For example, if a passenger exits the vehicle without fully closing a door, the robotaxi may refuse to proceed until the door is properly shut. Without a human present to latch the door, the car remains stuck with its hazard lights flashing, blocking traffic until someone intervenes. In many such cases, it falls to nearby pedestrians or passersby to take action, but people on the street are not a reliable or scalable on‑demand support network.

To address this gap, Waymo relies on a network of local workers who respond to assistance requests through an app similar to ride‑hailing platforms. Called by the robotaxi’s system when a simple fix is needed, these workers — often independent tow truck operators or gig workers — are paid modest fees to do tasks as basic as closing doors, unblocking seatbelts, or towing vehicles that have run out of charge. In cities like Los Angeles, these workers are typically offered around twenty dollars for closing a door and higher rates for physically towing the robotaxi to a charging station or depot.

The presence of this human workforce highlights a broader truth about automation: even as technology advances, machines still depend on people to bridge the gaps that software and sensors cannot yet cover. One such worker, a tow truck owner in the Los Angeles area, finds himself frequently summoned to rescue stranded robotaxis. He described reclaiming a vehicle by removing a seat belt trapped in the rear door — a mundane task that nonetheless highlights the kinds of unexpected “last‑mile” problems autonomous vehicles face.

The growth of the robotaxi fleet means these rescue tasks are becoming increasingly common. Waymo has announced plans to expand its services into multiple new cities, increasing the number of vehicles operating in diverse and often unpredictable urban conditions. With expansion comes the inevitable increase in human intervention requests — and also the need to cultivate a more robust support system that can respond quickly when a robotaxi runs into trouble. Some of the workers who respond through the assistance app report that locating stalled vehicles can be a challenge in itself, sometimes requiring them to walk through narrow side streets when the location data provided is insufficiently precise.

Another complicating factor highlighted by recent events has been large‑scale disruptions like power outages. When a blackout hit San Francisco, numerous autonomous vehicles found themselves immobilized at intersections because traffic lights were not functioning. Programmed to treat dark signals as four‑way stops, the robotaxis approached the intersections safely but then waited for confirmation from remote human supervisors to proceed. The resulting backlog of assistance requests caused gridlock as cars piled up, further underscoring how current autonomous systems still lean on human judgment during unusual circumstances.

The reliance on human support workers may seem like a step backward for a technology designed to eliminate the need for drivers altogether, but industry experts say it is a natural phase in the evolution of automation. As autonomous ride services scale up, the interaction between advanced machines and human labor is simply shifting. While early visions of driverless cars promised a world without drivers, the reality has become one where human workers fulfill new roles to ensure smooth operation. These roles may not have existed before — such as specifically closing robotaxi doors or retrieving stalled vehicles — but they nonetheless are essential to the service’s real‑world functioning.

Companies like Waymo are aware of these challenges and are experimenting with technological improvements to reduce dependence on manual intervention. For example, next‑generation robotaxi designs may include automated doors that no longer require human assistance, as well as improved battery management systems to prevent vehicles from stalling while in service. Training programs for riders, designed to educate customers on how to interact with autonomous vehicles, also aim to reduce user errors that lead to unnecessary intervention calls.

Even as automation advances, the job market around these services is evolving rather than disappearing. Workers who rescue robotaxis represent a new category of labor that did not exist before autonomous fleets took to the roads. Their work serves as a reminder that technology’s march forward often creates new opportunities for human labor even as it transforms whole industries. And in the near term, these human workers — summoned through apps and paid for simple but vital tasks — remain a crucial part of the infrastructure that keeps robotaxis moving smoothly through American cities.