Longtime partners Mobileye and STMicroelectronics are working on the next generation of Mobileye's chip that officials with both companies say will be used as the central processor for self-driving cars, a move that will increase their competition with the likes of Nvidia and NXP Semiconductors in the fast-growing market.
The same week that NXP unveiled its Bluebox autonomous vehicle platform, officials with Mobileye and STMicroelectronics outlined plans for the fifth generation of Mobileye's system on a chip (SoC), the EyeQ 5, which will collect and fuse the broad range of data being generated by the array of sensors on the vehicle.
The EyeQ5 will be designed on FinFET manufacturing technology at least 10 nanometers, if not smaller, and will feature eight multithreaded CPU cores and 18 cores of Mobileye's vision processors. The chip will deliver eight times the performance of the current EyeQ4 SoCs, produce more than 12 tera operations per second and will keep power consumption below 5 watts, which will enable the chips to maintain passive cooling, according to officials.
Mobileye is looking to leverage its capabilities around computer vision in vehicles to create the central processor that will fuse the data from the almost two dozen sensors in self-driving cars, such as cameras and radar to Lidar, speed and motion sensors.
"EyeQ5 is designed to serve as the central processor for future fully-autonomous driving for both the sheer computing density, which can handle around 20 high-resolution sensors and for increased functional safety," Amnon Shashua, co-founder, CTO and chairman of Mobileye, said in a statement, adding that the company is using its expertise and "our deep understanding of computer vision processing to develop highly optimized architectures to support extremely intensive computations at power levels below 5W to allow passive cooling in an automotive environment."
The autonomous vehicle market is expected to grow rapidly over the next several years as automakers and tech vendors expect to put self-driving cards on the road by 2020 or soon after. BMW officials earlier this month said they plan to have their first self-driving car, called the i Next, on the road by 2021. Analysts with Boston Consulting Group expect the market for partially and fully autonomous vehicles will reach $42 billion in 2025, and about $77 billion in 2035. ABI Research analysts in April said that the advanced driver assistance system (ADAS) market itself will grow from $11 billion this year to $132 billion by 2026.
A growing number of chip makers are making aggressive moves into the space. Nvidia is offering autonomous car platforms, including its Drive CX and Drive PX systems as well as its Driveworks software development kit (SDK). NXP, using the strength of its $12 billion acquisition last year of Freescale Semiconductor, is offering an autonomous car platform that puts its silicon into the Bluebox compute engine as well as into many of the sensors themselves.
Mobileye's EyeQ5 will feature four fully programmable accelerators in the chip that are optimized for their own domain-specific algorithms that will improve performance and power efficiency, and will include security capabilities that will support over-the-air software updates and secure in-vehicle communication, officials said. There also will be an SDK so automakers will be able to differentiate their offerings.
However, it will take a while for the EyeQ5 to get to market. Engineering samples aren't expected until the first half of 2018, with the full lineup of applications and the SDK coming in the second half of that year.