Real Roads Still Matter Most

Lou said the hardest problems involve everyday ambiguity, including hesitation, informal signals and partial commitments. A scooter cutting across a pickup point or a driver edging into a gap can test a robotaxi more than a clean, ideal-road scenario.

World Models Cannot Replace Exposure

The Pony AI executive argued that companies must move beyond supervised driver-assistance systems. He said a company only truly competes in the L4 market once it operates fully driverless vehicles across complex, scaled urban environments. In his view, commercialization depends as much on fleet operations as technology, because companies need enough real-world interactions to improve.

Lou said "world models" can help systems understand cause and effect from real-world data, such as how other road users react when a robotaxi slows or behaves cautiously at a junction. But he said those models cannot replace road exposure. “It is on the road, not in the lab, that the autonomous driving race will be won,” Lou states.

Robotaxi Race Keeps Getting Crowded

Pony AI, an artificial intelligence technology company focused on autonomous vehicles, says its PonyPilot service operates robotaxis in major Chinese cities. Its website describes PonyPilot as its robotaxi business, while the company has also developed L4 autonomous trucking technology.

According to Benzinga Edge Stock Rankings, Pony AI stock currently offers a poor Momentum score and an unfavorable Price Trend recommendation over Short, Medium and Long term.

Photo Courtesy: PJ McDonnell on Shutterstock.com