Is Baidu's L4 autonomous driving reliable by betting on AI chips?

In July 2017, Baidu CEO Robin Li sat in a self-driving car and drove on Beijing's Fifth Ring Road, which has one more ring than the Fourth Ring Road. Two news stories were posted on the same day: the first was that China's own self-driving technology has arrived, and the second was that it violated traffic rules and was fined.

In 2017, after repeatedly defining the group's new strategic direction as "All in AI", Baidu launched its first open autonomous driving system platform: Project Apollo. Simply put, this platform can collaborate with car companies to complete the development of autonomous driving for vehicles, and can also connect with hardware technology vendors to create an open platform in a bridging manner. At last year's Baidu World Conference, Robin Li made a promise: to complete mass production of driverless cars within a year.

Baidu went all in on AI and accomplished two major things in one year?

◆ Don’t look at the advertisement, look at the effect. What can L4 autonomous driving do?

◆ Betting on AI chips, how much strength can China's "core" Baidu add?

◆ Baidu has both autonomous driving systems and chips. Will the overseas market buy into it?

Time flies, and one year later, Lu Qi, the general manager of Baidu's intelligent driving business group, who had made a promise with Robin Li, resigned. But last week, Baidu still brought the self-driving mass-produced car they promised as scheduled, along with several eye-catching information: the world's first L4-level self-driving bus, which has achieved mass production capacity; it will be exported abroad for commercial operation; at the same time, China's first independently developed AI chip "Kunlun" was released, which beats its peers.

The news was exciting and people cheered. Many media reports on the same day mentioned that Baidu had made its boast come true. Baidu was the first company in the world to achieve mass production of driverless cars. China's chips were saved! But behind the news that swept the screen, some doubts also emerged: What can this L4 driverless car do? Can China's semiconductor industry, which lacks "chips", really succeed this time?

● What can Baidu’s L4 driverless bus do?

The L4 driverless bus that Baidu has put into mass production is called "Apollo". It is produced in cooperation with King Long Bus. The Apollo is 4.3 meters long and 2 meters wide. Since there is no driver's seat, there are 8 seats in the car, with a maximum capacity of 14 people (including 6 standing seats). It uses a pure electric power system and has a range of 100 kilometers. However, at this stage, the application of this car is limited to closed scenes, so don't expect to see it on the street at your doorstep.

In addition, through cooperation with Japan's SoftBank Group, this driverless bus will enter the Japanese market next year, but the application scenarios are still relatively single. According to Baidu's official statement, Apollon will be used in the Japanese market for personnel transfers within nuclear power plants and to serve some closed communities. Its main function is to replace shuttle buses.

Regarding production, Baidu officials mentioned that the current mass production volume has reached 100 vehicles, which is not a convincing number for the automotive industry. Moreover, from an objective point of view, the vehicles that have been offline are still in the state of "road testing" engineering vehicles, and even if they enter closed commercial operation, they will still need some time of testing. At the conference, Baidu did not give too much explanation in this regard.

Therefore, from the above two information points of application mode and mass production status, it can be seen that Baidu's L4 driverless bus, which has been mass-produced, is not a very amazing technical achievement in the industry. On the contrary, similar technical products have already had precedents in the United States and Japan. After all, driverless cars designed for closed or fixed route scenarios have completely different technical levels from vehicles that are actually used on open roads.

The technical requirements for autonomous driving technology on closed roads are different from those on open roads. First, the reliance on vehicle computing power, high-precision map support, image recognition rate, network base station and other technologies is lower. Second, it is easier to control vehicle speed and trajectory in a closed environment. Both aspects greatly reduce the technical difficulty of operation and achieve more sufficient safety redundancy. On the contrary, on open roads, the difficulty of each of the above technologies will increase exponentially.

Of course, due to the restrictions of laws and regulations, the application of autonomous driving on open roads has not been completely liberalized, which also proves the conservative attitude of the government and car companies towards this technology. Therefore, autonomous driving technology companies can only test it in closed places or compliant roads. In the face of autonomous driving technology that can really be put into open roads, there are still a lot of phased tests that need to be slowly overcome, and car companies are also very cautious in this regard. For example, Tesla is still delaying the update of the new version of Autopilot autonomous driving system.

From many dimensions, autonomous driving will face different technical challenges due to different application environments. However, as we mentioned above, it is still gratifying that Baidu has achieved a breakthrough in China's self-developed autonomous driving technology from scratch. However, as for "Apollo", it seems to be a bit of a big thunder and little rain, and it is more like "taking the L4 level out of context" to attract our attention. After the conference, some people complained online that the technical difficulty of Apollo, which has L4 level autonomous driving, is not even as high as the car that Li Yanhong took on the Fifth Ring Road last year.

As a winner of Toutiao's Qingyun Plan and Baijiahao's Bai+ Plan, the 2019 Baidu Digital Author of the Year, the Baijiahao's Most Popular Author in the Technology Field, the 2019 Sogou Technology and Culture Author, and the 2021 Baijiahao Quarterly Influential Creator, he has won many awards, including the 2013 Sohu Best Industry Media Person, the 2015 China New Media Entrepreneurship Competition Beijing Third Place, the 2015 Guangmang Experience Award, the 2015 China New Media Entrepreneurship Competition Finals Third Place, and the 2018 Baidu Dynamic Annual Powerful Celebrity.