Unpacking of Baidu’s car project

Article By : Junko Yoshida

The key to understanding Baidu’s automotive efforts is the company’s technical accumulation of AI and deep learning.

« Previously: Baidu enters robo-car derby with Project Apollo

What exactly do we know about Baidu’s robo-car project? Does it have the technical chops to grab the lead in autonomous driving?

According to Baidu, the Chinese Internet company established its Autonomous Driving Unit in 2016–based in Beijing and Sunnyvale, Calif., Baidu’s robo-car R&D, however, dates back much further.

Baidu's Institute of Deep Learning, one of three labs under Baidu Research, has led the autonomous driving project since 2013, according to the company. “Baidu AutoBrain, the core of Baidu's autonomous driving technology, includes highly automated driving (HAD) maps, positioning, detection, and smart decision-making and control,” Baidu said when the company concluded successful tests of self-driven vehicles in 2015.

The company said at that time:
The 30-kilometre test drive route begins at Baidu's Beijing Headquarters near Zhongguancun Science Park in Haidian District, extends to the G7 highway, Fifth Ring Road, Olympic Park, looping back and ending at Baidu's Headquarters.

The car demonstrated full autonomy on the entirety of the route and successfully executed driving actions including making right turns, left turns and U-turns, decelerating when detecting vehicles ahead, changing lanes, passing other cars and merging into traffic from on-ramps and exiting from off-ramps. The car speed peaked at 100km per hour during the test runs.

Baidu, which received an Autonomous Vehicle Testing Permit for California in September 2016, has done “multiple urban public road tests in a number of locations in China and California.”

The key to understanding Baidu’s automotive efforts is the company’s technical accumulation of AI and deep learning. Some of its technologies in autonomous driving include “environment sensing, behavior prediction, planning control, operation system, high precision localisation, high precision map and system safety,” explained Baidu.

Organisationally speaking, it’s notable that Baidu’s Autonomous Driving Unit is part of Baidu’s Intelligent Driving Group (IDG). Baidu says that IDG “combines Baidu’s Autonomous Driving Unit, Intelligent Vehicle Unit, and CarLife (Baidu’s telematics division) with a mission to accelerate auto industry upgrade in AI era.”

Baidu’s partners

So, who does Baidu already have as partners in the robo-car biz?

Bosch (radar), Harman (AI software), Nvidia (cloud-to-car platform) and Velodyne (lidar) come to mind, said Juliussen.

At one point, Baidu also had a joint research effort with BMW. This apparently fell through last fall, a few months after BMW-Intel-Mobileye alliance was announced. (When Baidu was carrying out road tests under complex road conditions in Beijing, the vehicle was a modified BMW 3 Series.)

Bosch announced earlier this year that it will collaborate with Baidu and China’s domestic mapping firms AutoNavi and NavInfo Co. on automated driving projects. The collaboration with Baidu, AutoNavi (owned by Alibaba Group Holding Ltd) and NavInfo (partly invested by Tencent) will “use Bosch radar and video sensors on cars to help generate and update maps critical to autonomous driving capabilities,” according to Bosch.

Meanwhile, at the Shanghai Auto Show last week, tier one Harman–now owned by Samsung–announced a new partnership with Baidu. This union is expected to bring together Harman’s broader reach among automakers, and its expertise in-car infotainment systems, with Baidu’s AI software.

According to the announcement, Baidu will develop cloud-based AI solutions for use in cars with Harman, across efforts that include building Baidu’s DuerOS voice-based personal assistant into cars.

Baidu’s collaboration with Nvidia is centered on AI, developing a cloud-to-car autonomous vehicle platform for local Chinese and global car makers.

Baidu_autonomous-car_02 (cr)
Figure 1: Nvidia announced last summer partnership with Baidu on Map-to-Car Platform for Self-Driving Cars (Photo: Nvidia)

In the announcement at Beijing’s Baidu World Conference last summer, the two companies explained that Baidu’s cloud platform and mapping technology will combine with NVIDIA’s self-driving computing platform. The goal is to develop solutions for HD maps, Level 3 vehicle control and automated parking.

Meanwhile, Baidu invested $75 million in Velodyne last year, along with Ford—which also made $75 million investment in the lidar maker.

Next: Why Baidu needs high-profile car OEM deals »

Subscribe to Newsletter

Leave a comment