Tesla quietly launched its first robotaxi service in Austin, Texas, on Sunday, deploying a pilot fleet of about 10 Model Y SUVs. The fare is $4.20, a nod to Elon Musk's announcement on X (formerly Twitter). Several influencers have already posted videos showing the vehicles in service in the city.

While this rollout marks a long-awaited turning point, Tesla remains cautious: each car has a "safety monitor" in the front passenger seat, and a remote control system is activated behind the scenes. This technique, known as teleoperation, is increasingly used in the autonomous vehicle industry. Here's what we know about how it works and its limitations.

What is teleoperation?

Teleoperation allows a human to control a vehicle or robot remotely via a wireless network. In the case of robotaxis, it is mainly used to monitor the smooth running of journeys, assist artificial intelligence in ambiguous situations or, as a last resort, temporarily take control of the vehicle.

How is it used in industry?

At Waymo (an Alphabet subsidiary), human agents called "fleet response" can be called upon by the vehicle itself. Human intervention is limited to providing contextual information, without direct control of the vehicle, which remains the "final decision-maker," as former CEO John Krafcik explained.

Conversely, Baidu in China has tested a more radical model, with backup drivers capable of remotely controlling the vehicle. However, these solutions are still rare on a large scale.

What are the limitations?

One of the major risks of remote operation is dependence on the quality of the network connection. A signal interruption or delay could prevent a remote operator from responding effectively to a critical situation.

Philip Koopman, an expert in autonomous vehicle safety at Carnegie Mellon University, believes that this technology may be suitable for a fleet of ten vehicles, but that it is "fundamentally unreliable" on a large scale: "With a million vehicles, it will happen every day," he warns.

Furthermore, a single remote operator cannot monitor a large number of vehicles simultaneously without compromising safety. Finally, letting AI make decisions alone in complex situations remains risky: it could misinterpret a situation or choose the wrong course of action.

What is Tesla's strategy?

For years, Elon Musk has been promising the transition to fully autonomous driving via his Full Self-Driving (Supervised) software. In January, he announced the arrival of "unsupervised" robotaxis in Austin for June. But the current launch remains highly restricted: limited traffic areas, complex intersections avoided, and remote operators ready to intervene.

Tesla has published job offers for teleoperation positions, specifying that these employees will be able to remotely "access and control" autonomous vehicles or humanoid robots. Their mission: to perform complex maneuvers remotely if necessary.

A Tesla employee says the strategy calls for human intervention in the event of unforeseen circumstances, such as a robotaxi getting stuck in the middle of a crowd of pedestrians.

Increased political vigilance

This launch comes despite calls from several Texas Democratic lawmakers asking Tesla to postpone the initiative until September, when a new law on autonomous driving comes into force. They cited public safety concerns and trust issues with the system.

Tesla did not respond to requests for comment, but Musk insists that the company is "extremely cautious" on safety issues.

For now, this project remains a small-scale test. But it represents a new step in the race toward fully autonomous vehicles. The question remains whether teleoperation, as a technological crutch, can withstand the shock of mass deployment.