The Chinese internet firm Baidu, well renowned for its search engines, is making significant progress in autonomous driving.

The company’s robotaxis in Wuhan will begin operating without human safety drivers between the hours of 7 a.m. and 11 p.m. The previous city restrictions limited the time of day its autonomous vehicles could operate to 9-5. Wuhan is home to about 10 million people, and the modified plan is projected to serve 1 million residents there.

Unlike Tesla’s vision-based approach, Baidu’s autonomous vehicles use a combination of third-party cameras, radars, and lidars to improve visibility in low-light environments.

In August, Baidu began providing rides in fully autonomous robotaxis at taxi fares. More than 474,000 rides were completed in Q3 using the company’s robotaxi-hailing app, Apollo Go, which is a 311% year-over-year increase. As of the third quarter, the total number of orders for Apollo Go had surpassed 1.4 million.

It’s reasonable to question how many of those trips are being subsidised by discounts, but it still seems like a sizable cash stream for Baidu. When compared to the novelty trips taken by early adopters, what percentage of these are regular, everyday routes? It is typical practise in China for robotaxi companies to actively solicit customers to increase their passenger counts and hence their profits.

It’s also unclear at this time which of the several Chinese robotaxi startups is now in the lead. Because of the importance of their business relationships with the cities in which they operate, and because big cities typically have the authority to approve specific municipal legislation, their growth is dependent on such relationships.

The autonomous driving industry is receiving enthusiastic backing from local governments around the country since it is one of the few consumer internet industries with significant growth potential. Industrial Wuhan in central China is one of the first places in the country to allow robotaxis to transport the public without human safety drivers. The city’s residents are increasingly used to seeing autonomous automobiles out and about, even in dusk and dawn.

Putting aside some healthy scepticism, Baidu has made significant progress in bringing the autonomous driving future forward in time. It is fortifying its position with a visual-language model that can detect and name uncommon items in long-tail circumstances. Wenxin, the same huge model that supports its text-to-image art platform, is the AI’s backbone.

“The model will enable autonomous vehicles to quickly make sense of an unseen object, such as special vehicle (fire truck, ambulance) recognition, plastic bag misdetection, and others,” Baidu previously explained. “In addition, Baidu’s autonomous driving perception model—a sub-model of the WenXin Big Model—leveraging more than 1 billion parameters, is able to dramatically improve the generalization potential of autonomous driving perception.