Skip to Content

The Self-Driving Project That Could Help China Leapfrog the West

Baidu opens up its software, a stark departure from the normally secretive world of commercial AI development.
One of Baidu’s experimental self-driving cars.
Apollo

The CEO of Baidu, Robin Li, arrived at his company’s first AI developer conference, held in Beijing this week, in a vehicle that has the potential to reshape the world of self-driving cars.

The vehicle was controlled using software that Baidu (50 Smartest Companies 2017) plans to offer for free in the coming years through a project called Apollo. By making the brains of a self-driving car available to anyone, the Apollo project could help China’s many young carmakers get up to speed rapidly.

It also reflects China’s broader ambition to establish itself as a leading hub of artificial intelligence. Baidu’s move of making its training data openly available marks a significant departure in the field of commercial AI, where the information used to train sophisticated algorithms is typically guarded with obsessive jealousy.

This fits with the Chinese government’s desire to see its nascent AI industry become increasingly competitive, and to see the underlying technology feed into its planning for the future, including the design of new cities. Indeed, if the Apollo effort gains momentum, it may make it harder for any of the companies protecting their own code to dominate the automated driving field.

“Apollo is an important milestone for the automotive industry,” Qi Lu, vice chairman of Baidu, told conference attendees. “It is in essence the Android of the autonomous driving industry, but more open and more powerful.”

The time line for developing autonomous cars using the technology is ambitious. The goal is to begin testing Apollo vehicles on restricted areas later this month and to have fully autonomous driving on urban roads and highways by the end of 2020.

The Apollo platform certainly seems to be effective for bootstrapping the development of self-driving vehicles. At the event in Beijing, a California-based startup called AutonomousStuff demonstrated a Lincoln that it turned into a rudimentary self-driving car using the Apollo technology in just three days.

The Apollo platform consists of a core software stack, a number of cloud services, and self-driving vehicle hardware such as GPS, cameras, lidar, and radar.

The software currently available to outside developers is relatively simple: it can record the behavior of a car being driven by a person and then play that back in autonomous mode. This November, the company plans to release perception capabilities that will allow Apollo cars to identify objects in their vicinity. This will be followed by planning and localization capabilities, and a driver interface.

The cloud services being developed by Baidu include mapping services, a simulation platform, a security framework, and Baidu’s DuerOS voice-interface technology.

Dawen Zhou, Apollo’s principle product manager, explained that the simulation platform being developed by Baidu would be used to test the code and also to train self-driving algorithms. Since lots of real-world driving data is vital for the continued improvement for Baidu's self-driving technology, one of the biggest benefits of opening up the Apollo platform is the data Baidu can receive from its partners. "This is the real treasure of Apollo," Zhou said.

At the AI conference, Baidu and the U.S. chipmaker Nvidia (50 Smartest Companies 2017) announced plans to work together on several AI initiatives. This will include using Nvidia’s Drive PX platform as part of the Apollo project.

The Apollo effort has already attracted an impressive number of partners. A number of Chinese automakers are signed up, including Chery Automobile, Dongfeng Motor, Foton, Yiqi, and FAW Group. Some major European and U.S. companies, including Bosch, Ford, Intel, Microsoft, and Velodyne, have also agreed to contribute. Other partners include major Chinese universities and government bodies.

Stephen Zoepf, executive director of the Center for Automotive Research at Stanford, says the Apollo approach could prove helpful in persuading the authorities to accept the technology. “A certain degree of openness will be important for buy-in in any of these projects,” says Zoepf. “Perhaps not to the extent of complete open-source, but at least to the extent of public and regulators to have a sense of the level of risk involved.”

With reporting from Yiting Sun in Beijing.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.