Tesla's robotaxi ambitions facing a bumpy ride

Tesla finally has a robotaxi. Now comes the hard part.

The electric vehicle maker deployed its first-ever driverless cabs in Austin, Texas, on June 22 in a small-scale test of carefully monitored Model Y SUVs. Next, the company faces the steep challenge of delivering on CEO Elon Musk's ambition to refine the software and upload it to millions of Teslas within a year or so.

Such a rapid expansion will prove extremely difficult, said industry analysts and autonomous vehicle technology experts. These observers expressed a range of views about Tesla's prospects but all cautioned against assuming a light-speed robotaxi rollout.

Some pointed to advantages Tesla might exploit to overtake rivals including Alphabet's Waymo and a host of Chinese auto and tech companies. Tesla has mass-manufacturing capacity, and it pioneered remote software updates it can use for self-driving upgrades. The automaker also does not use sensors such as radar and lidar like Waymo and most rivals; instead, it depends solely on cameras and artificial intelligence.

"A rollout could be really quick. If the software works, Tesla robotaxi could drive any road in the world," said Seth Goldstein, Morningstar strategist, while cautioning that Tesla is still "testing the product".

In Austin, Tesla launched a choreographed experiment involving maybe a dozen cars, operating in limited geography, with safety monitors in the front passenger seat; remote "teleoperators"; plans to avoid bad weather; and hand-picked pro-Tesla influencers as passengers.

For years, Musk has said Tesla would soon operate its own autonomous ride-hailing service and also turn any Tesla, new or used, into a cash-generating robotaxi for its customers. That will be "orders of magnitude" more difficult than testing in Austin, said Bryant Walker Smith, a law professor focused on autonomous-driving regulation at University of South Carolina.

"It's like announcing that, 'I'm going to Mars' and then, you know, going to Cleveland," Smith said.

Musk has said Tesla will reach Mars, using that metaphor, quite quickly: "I predict that there will be millions of Teslas operating fully autonomously in the second half of next year," he said in April.

Musk and Tesla did not respond to requests for comment.

Given Tesla's AI-dependent approach, its challenge will be machine-training robotaxis to handle complex traffic "edge cases", said Philip Koopman, a computer-engineering professor at Carnegie Mellon University and autonomous-technology expert. That could take many years.

"Look, how long has it taken Waymo?" Koopman asked. "There's no reason to believe Tesla will be any faster."

Waymo's self-driving efforts date back to 2009, when Google started its self-driving car project. An egg-shaped prototype took its first ride on public streets in 2015, also in Austin.

It has taken Waymo since then to build a 1,500-robotaxi fleet in select cities. A Waymo spokesperson said it plans to add 2,000 more vehicles by the end of 2026.

Some analysts believe Tesla can expand faster, in part because Waymo has helped pave the way by overcoming regulatory and technical challenges.

"Waymo and other pioneers have helped to drive regulatory change and have made riders, pedestrians and other road users aware of autonomous vehicles," said Paul Miller, an analyst at market-research firm Forrester.

Reuters