domeet webmaster

On the fourth-quarter performance conference call in 2020, Xiaopeng Motors CEO He Xiaopeng said, “In 2021, we will increase R&D investment, including hardware closely related to autonomous driving.”

In North America, the lead of the chip project is Benny Katibian, COO of Xpeng Motors North America, and the head of the domestic chip project is Xia Heng, co-president of Xpeng Motors. The above-mentioned industry insiders said that chip research and development is a complex project, and it is normal to carry out multiple locations.

Benny Katibian, like Wu Xinzhou, vice president of Xiaopeng Auto’s autonomous driving business, is from the chip giant Qualcomm. After joining Xiaopeng Motors, he is responsible for hardware, architecture and system design at the Autonomous Driving Center, while maintaining Xiaopeng Motors’ and Nvidia’s, Cooperative relationship with chip companies such as Qualcomm.

In China, Xiaopeng Motors is also actively recruiting chip technicians. According to industry insiders, some domestic chip engineers have received an invitation from Xiaopeng Motors, “The position requires full-process management of the chip project from project establishment to implementation (deployment).”

A few days ago, when the CEO of Xiaopeng Motors was interviewed by other media, he responded to whether he had self-developed chips, saying, “If I do, I will tell you about it.”

Tesla is the “first person” self-developed self-driving chips for automakers, and the vertical integration of software and hardware makes its smart technology lead all the way. The leading domestic new-built car companies are consistent with Tesla’s self-driving research and development strategy, all pursuing full stack self-research. “Tesla is done, everyone just copy the homework.” An executive of a new car company commented.

Currently, Weilai, Xiaopeng Automobile and Ideal Automobile have all chosen to independently research and develop software and algorithms. Xiaopeng Motors started to build a self-research team in 2017, and the largest self-research result-the high-speed autonomous driving navigation function NGP has been delivered. During the difficult period of 2019, NIO once shelved the R&D progress in North America. After recruiting the former Momenta R&D director Ren Shaoqing, it restarted in China. The autonomous driving self-research system of Ideal Auto will also be promoted and improved by CTO Wang Kai in the second half of 2020.

Focusing on self-developed software, the three companies have chosen Nvidia’s Orin chip for their autonomous driving platforms. The latter has a computing power of 200 TOPS and a power consumption of 45W.

But the use of third-party computing chips is not in conflict with self-research, and walking on two legs has been practiced by Tesla. At the beginning of 2016, former AMD chief architect Jim Keller joined Tesla as the vice president of Autopilot. Later, Jim Keller’s friend Peter Bannon also chose to join. Tesla’s self-developed chip project was officially launched. By April 2019, it was officially in It is installed on mass-produced models. In this process, Tesla’s autopilot chips have successively adopted Mobileye’s EyeQ3, NVIDIA Drive PX2 and other products, until the results of its own self-developed chips are implemented.

I learned that the three domestic leading car companies have also been involved in the research and development of autonomous driving hardware to varying degrees.

In October last year, there was an exclusive report that Weilai CEO Li Bin was actively promoting the self-developed chip project and also recruited Bai Jian, the former general manager of Xiaomi’s chip business, as the vice president of hardware business. At the Weilai annual conference in January, Li Bin implicitly responded to the media about his self-developed chip plan, “Autonomous driving chips are not as difficult as general-purpose chips. We will maintain our core competitiveness.”

It is not clear whether the ideal car is self-developed or not. The company’s CTO Wang Kai once said that the logic of self-developed special-purpose chips is to first run the algorithm. Nvidia has a mature tool chain that can help car companies quickly establish self-development. Research algorithm. However, it was recently learned that Wang Kai’s department has set up a cutting-edge technology laboratory, and it is not clear whether it includes pre-research on chips.

Xpeng Motors, which uses smart technology as its banner, has more positions to promote self-research of chips. The company’s software system is relatively mature and has clear requirements and definitions for autonomous driving hardware. Xiaopeng Motors CEO He Xiaopeng also told that he has participated in many hardware-level designs, including the design of embedded boards. For Lidar, “not only used, but also required for our length, width, height, performance, Many functions have been adjusted”.

Some hardware engineers commented that one of the benefits of self-developed chips is to make the best use of everything. “It can maximize performance and stability. Compared with general-purpose chips, it can also remove some unused things and reduce costs. .”

The Xiaomi Group, which just announced the creation of a car, also released the image processing chip Surging C1. In an interview with CCTV, the founder of Xiaomi Group Lei Jun said bluntly, “The hardware industry has a lot of technical thresholds and technology accumulation, and finally it is in the form of chips. To reflect.”

The technical armament race has long been the norm in the new car camp, and it seems that it is only a matter of time to get involved in chip research and development, depending on whose pace is faster.

Extended reading:

Focus analysis | Spend tens of billions of dollars, but Xiaomi may notBuilding “the first car for young people”

Exclusive | Xiaomi car manufacturing project is expected to be established in April, Wang Chuan will be in charge, and the auto industry is being explored