Car stuff (public number: chedongxi)
The author | James
Edit the | Xiao Han
Car East January 5 news, Beijing time in the early morning of this morning, CES 2022 (International Consumer Electronics Show) officially kicked off. In the field of autonomous driving, Intel subsidiary Mobileye released three self-driving chips, one of which is the EyeQ Ultra chip for L4 autonomous driving, and the other two are EyeQ6L and EyeQ6H chips for L2 automatic driving, and announced the latest cooperation progress.
Among them, EyeQ Ultra is a higher computing power of the automatic driving chip, based on the 5nm process, its AI performance can reach 176TOPS, about 10 EyeQ5 chip performance. The EyeQ Ultra has a 12-core, 24-thread CPU, two general-purpose compute accelerators and two CNN accelerators. The EyeQ Ultra is expected to be available in samples in 2023 and in mass production in 2025.
EyeQ6H is an iterative version of the latest mass-produced chip EyeQ5, the computing power is equivalent to 2 EyeQ5s, about 50TOPS; EyeQ6L is an iterative version of the previous generation of L2-level autonomous driving chip EyeQ4, but the package size is 55% of EyeQ4. Both chips in the EyeQ6 series will be built based on the 7nm process, and samples are expected to begin this year and mass production will be achieved in 2024.
In addition, Mobileye announced a series of partnership plans. Mobileye and Extreme Krypton's L4 self-driving model will be mass-produced as early as 2024, and its autonomous driving system will be based on six Mobileye EyeQ5s. Because this is a consumer-grade product, it means that the average user can buy it directly. At the same time, Mobileye has also reached in-depth cooperation with Volkswagen Group and Ford Motor.
As the world's largest supplier of L2-level autonomous driving chips, Mobileye is also expanding its L4 autonomous driving business in addition to holding the L2-level autonomous driving position. In the context of increasingly fierce competition for autonomous driving around the world, Mobileye also hopes to catch up with competitors such as Nvidia and seize the key opportunity to build the future of mobility.
First, the launch of three automatic driving chips with a maximum computing power of 176TOPS
Mobileye's EyeQ Ultra for L4 autonomous driving is the chip with the highest computing power in Mobileye's history, built using a 5nm process, and its AI computing power can reach 176TOPS.
▲EyeQ Ultra architecture diagram
According to the chip architecture diagram published by Mobileye, the EyeQ Ultra has a 12-core 24-thread CPU, as well as GPUs, VPU, ISP units and security cores. In addition, there are four compute acceleration units, two of which are general-purpose compute acceleration units and the other two are CNN acceleration units. The chip also has an LPDDR5x channel and a fast communication and video signal interface.
According to the introduction, EyeQ Ultra can be paired with CPUs, ISPs, and GPUs to achieve more efficient computing. The result is the ability to process perceptual signal inputs from cameras, millimeter-wave radar, and lidar.
In terms of computing power, the EyeQ Ultra's 176TOPS hashrate is equivalent to 10 EyeQ5 chips. Mobileye says it can handle the needs and applications of L4 autonomous driving with a single chip, without the need to integrate multiple SoCs to increase power consumption and cost.
Another Mobileye EyeQ6H chip positioning L2 autonomous driving is an iterative product of the current EyeQ5, which is based on a 7nm process and has the same computing performance as 2 EyeQ5s.
The EyeQ6H chip has a similar architecture to the EyeQ Ultra, with an 8-core 32-thread CPU, but lacks VPU units, and the specifications of the General Purpose Compute Acceleration Unit and the CNN Acceleration Unit are also reduced.
▲EyeQ6H chip architecture
This chip solution will provide richer L2+ ADAS capabilities and enable multi-camera signal processing as well as panoramic imagery and driver monitoring. The chip is expected to be available in samples this year and mass production will begin by the end of 2024.
Finally, there is an entry-level EyeQ6L chip, which is also based on the 7nm process and will be introduced to the market as a successor product to the EyeQ4 chip. Its CPU is dual-core 8-thread, which reduces the GPU and ISP unit compared to EyeQ6H, and the general computing acceleration unit and CNN acceleration unit are reduced by one each, and the specifications are reduced.
▲EyeQ6L chip architecture
The chip will be similar to the EyeQ4's packaging scheme and will be mounted directly on the inside of the vehicle's windshield. The EyeQ6L chip was already available to customers last year and will begin mass production in mid-2023.
Second, expand the cooperative relationship will be with the extreme krypton to build L4 autonomous vehicles
In addition to launching new chips, Mobileye has also reached deeper cooperation with a number of automakers, the most important of which is the creation of consumer-grade L4 autonomous driving models with Extreme Krypton Automobile.
According to reports, mobileye and krypton to build a self-driving car will be equipped with six EyeQ5 self-driving chips, which will be used to support Mobileye's true Redundancy sensing system, driving strategy based on responsibility-sensitive safety model (RSS) and open collaboration mode based on road network information management (REM) mapping technology.
Mobileye has partnered with Extreme Kr to develop consumer-grade autonomous vehicles
The model will be built on the Geely SEA's vast architecture, with redundant braking, steering and power functions all being done by Kr. At the same time, under the open EyeQ concept, mobileye and krypton software technology are efficiently integrated.
According to the planning of the two sides, the first L4 autonomous driving model jointly created by the cooperation will be launched in China as early as 2024, and it may become the world's first consumer model with L4 level automatic driving capabilities. In the future, this car will also be gradually introduced to the global market.
In addition, Mobileye announced that it will strengthen its R&D capabilities in China, establish a local data center and expand Mobileye's China team to support the fast-growing Business in China.
In the field of L2 autonomous driving, Mobileye has expanded its partnership with volkswagen group and Ford motor vehicles.
▲Mobileye has reached many cooperation with car companies
At present, the Volkswagen Group is using mobile RoadBook to enhance the driving experience of the Travel Assist 2.5 system. For example, in some road sections, vehicles are able to provide lane keeping assistance in areas without obvious lane markings. Mobileye's proprietary Road Information Management (REM) technology automatically aggregates and generates autonomous vehicle maps in the cloud, providing a truly global and scalable mapping solution for autonomous vehicles.
Volkswagen, Skoda and Seat electric vehicles based on the Volkswagen MEB platform will soon launch the Travel Assist function equipped with RoadBook.
In addition, Ford also announced the expansion of cooperation with Mobileye, Ford will begin to use Mobileye's road information management (REM) self-driving car mapping technology in future versions of Ford's BlueCruise L2 self-driving platform, which can allow drivers to drive off their hands while L2 autonomous driving is turned on.
Based on REM technology, Ford's existing adaptive cruise control (ACC) will also add automatic start-stop control and lane centering (LCC) functions. The two sides will also collaborate on Mobileye's open platform, allowing Ford to build and integrate its own solutions to make driving safer and easier in the future.
Conclusion: Mobileye is working on the future of mobility
As the world's largest supplier of autonomous driving chips, mobileye has ushered in a surge in shipments, but it has encountered many new forces. Therefore, although Mobileye is still the industry leader, it is still very anxious in the face of future travel.
Today, when L4 level automatic driving is moving towards commercialization, Mobileye is making L4 autonomous driving, that is, laying out foreseeable future travel scenarios, which is enabling this veteran autonomous driving company to achieve take-off in the field of future mobility.