
Tesla Robotaxi deadline is approaching, but "key tests" have not been conducted yet

With only a few weeks left until Elon Musk's promised launch of the first Robotaxi service in Austin, Texas, Tesla has yet to begin testing its unmanned safety drivers as of last month. The starkly different training and operational paths compared to Waymo have also raised concerns about the safety of Tesla's Robotaxi
Tesla Robotaxi's Final Sprint: Autonomous Driving Tests Not Completed, Uncertainty Remains for Austin Deployment.
According to media reports, there are only a few weeks left until Musk's promised deadline to launch the first Robotaxi service in Austin, Texas, but as of last month, Tesla has not yet begun testing with a human safety driver—this is a critical step before launching a pilot service to customers.
Internal emails show that Tesla has been training the Model Y in Austin since last fall. In April of this year, the company stated that employees were testing a "supervised" Robotaxi service equipped with human safety drivers. However, when the autonomous driving testing phase will begin remains unclear.
This is crucial for Tesla and its investors. Musk is betting the company's future on Robotaxi, which will initially rely on existing Tesla vehicles, followed by a dedicated Cybercab that is planned to begin production next year.
Meanwhile, Waymo, a subsidiary of Alphabet, has been operating commercial Robotaxi services in multiple U.S. cities, using a completely different training and operational approach.
Waymo vs Tesla: Two Distinct Training Paths
When entering new cities, Waymo uses sensor data from a small number of test vehicles to prepare detailed maps to ensure safe navigation for the vehicles. This process took ten years to develop and requires at least a year to complete.
In Austin, Waymo conducted about six months of testing with safety drivers in the vehicles, followed by six months of testing without safety drivers.
Currently, Waymo vehicles have not transported passengers on the major thoroughfare I-35 in Austin and are still training on highways in California.
In contrast, Tesla primarily relies on data accumulated from existing Tesla owners to train its cars.
An engineer close to the testing told the media that as the company prepares to launch the service in Austin, it faces issues such as rerouting when cars get stuck and providing customer service—essentially in situations where no one is in the car.
Waymo and other competitors have spent years addressing these issues.
Tesla's "Camera Dependence" Raises Safety Concerns
Unlike Waymo, which uses laser and radar systems, Tesla relies solely on cameras. Critics argue that this makes the latter's autonomous driving approach more prone to errors.
Missy Cummings, a researcher in autonomous vehicles at George Mason University, bluntly stated:
"Tesla's approach simply does not work. Computer vision is just not sufficient for this task."
This month, Musk told investors that the Austin launch may only include 10 Model Y vehicles. The company is hiring a remote team to supervise cars that encounter problems using VR headsets.
Many observers expect these vehicles to operate only in limited areas of the city. Tesla has not indicated whether the service will include the I-35 highway
The Performance of FSD Remains to Be Observed
Reports indicate that Tesla's automated driving assistance service may also have safety risks.
Tesla's annual safety report released last month shows that drivers using Autopilot (its advanced cruise control system) have a collision rate that is 10 times lower than the national average—however, this data is from highway driving. Other companies that have launched Robotaxi services in urban areas, including Uber and Cruise, have lost years of progress due to single collision incidents.
According to data from a crowdsourced tracking website, Tesla drivers using the semi-automated FSD mode can average 238 miles on city streets before needing to take control of the steering wheel.
However, based on investigations by federal regulators into two incidents in 2023, Autopilot previously failed to prevent collisions with obvious obstacles, such as children near school buses and the sides of tractor-trailers.
During the earnings call in April this year, Musk stated:
"The ultimate test is whether you can sleep in the car and then wake up at your destination. I believe that by the end of this year, many cities in the U.S. will offer this service."
However, he had previously promised that Tesla would reach the same benchmark by the end of 2020