/Baidu’s vision for autonomous driving

Baidu’s vision for autonomous driving

Baidu Senior VP and General manager of Baidu’s Autonomous Driving Unit (ADU), Wang Jing, introduced on Thursday Baidu’s autonomous driving goals at the 2016 Baidu Technology Innovation Conference in Beijing.

“With autonomous driving, we believe that urban transport can become safer, more efficient, and more economical,” said Wang.

Wang speaking at the conference (photo taken by Ke Jin)
Wang speaking at the conference (photo taken by Ke Jin)

By introducing the idea of software defined vehicles (SDV), Baidu believes that the value of cars in the future will depend mostly upon the software, not the car itself. By employing thousands of GPU, and more servers than what was used by Google’s AlphaGO for cloud computing and deep learning, a model will be built. This model will be downloaded to the autonomous vehicles, data from which would then be uploaded to the cloud for further learning. Wang believes that autonomous cars must be tens or hundreds times safer than cars driven by humans.

According to the Beijing Transportation Research Center (BJTRC), Beijing commuters who drive or take buses and taxis are estimated to waste nearly two hours per day trapped in traffic jams. Wang said that, since autonomous vehicles do not cut into traffic as humans often do, they can effectively avoid the chaos and inefficiency that occurs when traffic jams become ubiquitous.

Baidu also expect that Autonomous vehicles will make transport more economical. Though the lidar device – a crucial navigation component for autonomous driving vehicles – on the top of Baidu’s autonomous car at present costs USD 80,000, Wang said that the price may be reduced to USD 500 if a much larger order is placed. Baidu expects to largely reduce the manufacturing cost of an autonomous car within five years.

Baidu's autonomous car exhibited at the conference (Photo taken by Ke Jin)
Baidu’s autonomous car exhibited at the conference (Photo taken by Ke Jin)

With a world-leading vehicle recognition rate of 90.13%, Baidu’s autonomous car can now read traffic lights and run on urban roads. It had its first road test in last December and is now working to get its “driver’s license“.

The Baidu ADU currently has four R&D centers, including a team of 100 engineers in Silicon Valley, the major center in Beijing, and the other two in Shanghai and Shenzhen. In addition, other than attracting talent from around the globe, Baidu has also opened their platform to more than one million developers and over 600,000 enterprises.

With attributions to Baidu’s achievements in autonomous driving and voice recognition, Baidu was recently ranked by MIT Technology Review as one of the 50 smartest companies in 2016, second only to Amazon. Additionally, Baidu received on Thursday the world’s 15th Autonomous Vehicle Testing Permit from the California Department of Motor Vehicles, allowing Baidu to test its autonomous driving technologies in the U.S. according to Wang.

Baidu will work with parties from all sides including central and local governments, sensor makers, auto makers, and research institutes.

(Top photo from Baidu Images)

Ke graduated from the University of Edinburgh with a Master's Degree in English and has worked on projects with Ipsos MORI and SDI Media. She's particularly intrigued by China's thriving technology scene and is eager to write about this flourishing industry.