Baidu's announcement on Project Apollo—the Chinese Internet behemoth's plan to "open up" its autonomous driving platform to others—has surprised many auto industry watchers, largely because autonomous car development remains embryonic.

However, given Baidu’s heavy R&D investment in artificial intelligence technologies and the many business deals the company has made in the auto industry, it’s understandable that Baidu wants to take control of robo-car software. Accomplishing this objective by opening its platform is both ambitious and, most agree, a smart move.

Phil Magney, founder and principal advisor for Vision Systems Intelligence, described the Baidu plan as “a bit of a moon shot.” But China is the right place to try this out, he noted. “The Chinese OEMs don’t have a lot to lose with this and it gives them the technology to jump in quickly.”

Baidu_autonomous-car_01 (cr) Figure 1: Baidu's autonomous car during test drive in China.

At this point, Baidu hasn’t disclosed details of Project Apollo. Interpreting what it entails boils down to educated guessing among analysts in the automotive and tech industries.

Mike Demler, senior analyst at The Linley Group, told EE Times, “I would expect [Baidu] to follow an ‘open’ model similar to Google’s Android operating system.”

In an “open” model, Baidu would control the OS, allowing others to add apps on top. “Baidu will tie Apollo to their autonomous-driving cloud services, as well as other non-driving related apps, similar to how Google generates revenue from their ‘free’ OS,” Demler explained.

Egil Juliussen, director of research for infotainment and ADAS at IHS Automotive, goes a step further. He sees Baidu’s move as China’s attempt to set a standard API for highly automated vehicles. By opening Baidu’s vehicle software platform to others, “Baidu is providing a lot of APIs. API for driving engines, steering wheels, API for sensors and sensor fusion, API for stored map information, etc.,” he explained.

In short, APIs become the key for Baidu to build an ecosystem, noted Juliussen. As Baidu receives more input from developers and more vendors design applications, Baidu will be in the catbird seat to improve and control APIs.

One thing that everyone agrees on is that “the Baidu plan is not about building a car or anything to do with it,” as Magney put it.

In Magney’s view, Baidu’s Apollo “sounds like a reference design and software development kit.” He suspects, “It is completely open from the sounds of it. Certain elements of the code base for automation will be open source.”

Search is the past

For Google and Baidu, the end game isn’t a car. But getting into cars will help them expand big data and advance their power in artificial intelligence. A narrative common to both Internet search giants is that traditional search is the past and artificial intelligence is the future.

But here’s an intriguing question. Why is Baidu opening its autonomous vehicle software platform now, while Google/Waymo stands pat?

Magney noted, “Even though Waymo shows limited evidence of doing what Baidu is doing, I believe Waymo’s long term end game is just that. Controlling the internet of cars. For both companies, it’s all about mobile devices, and in this case the car is another device.”

Most analysts unanimously believe Waymo is already far ahead of Baidu when it comes to autonomous vehicles. IHS Automotive’s Juliussen said, “When you are way ahead [like Waymo is], why would you want share it with anyone?”

The Linley Group’s Demler agreed. “Technically I expect Waymo is more ready to hit the road than Baidu. After boasting that we’d have self-driving cars by 2020, I think Waymo is being sensible by continuing to test, test, and test some more as they refine their system,” said Demler.  
Next: Unpacking of Baidu's car project »