STMicroelectronics’ EyeQ5: Too late in the game?

Article By : Vivek Nanda

Last month, STMicroelectronics announced a 2018 availability of the EyeQ5. Are they too late in the game?

Last month, STMicroelectronics announced that together with Mobileye, a vision processor core company with which they have been collaborating since 2004, they are developing the next generation of Mobileye's SoC for autonomous driving.

What's special about the device, according to STMicroelectronics, is that the EyeQ5 will use the 10nm or below FinFET technology node and will feature 8 multithreaded CPU cores and 18 Mobileye vision processing cores. They plan using these enhancements to target a performance increase of 8x over EyeQ4, run over 12 Tera operations/s, keep power consumption below 5W and maintain passive cooling.

This is being seen by market analysts as a "pre-emptive strike against NXP’s Bluebox." In an email exchange with EE Times, Elchanan Rushinek, Mobileye’s senior vice president of engineering, said that the future holds 360-degree coverage by cameras and ultrasonic, with part of it covered by radar/LIDAR. He sees some dedicated processors for specific sensors and central processors for fusion and decision making; the latter function being performed by vision processors [18 in the EyeQ5].
In that interview, Rushinek explained, “The nature of central processing is heavily around perception of the scene, which makes vision processors, like EyeQ4/5, the natural choice for the central processing.”

STMicroelectronics is not alone in seeing a bright future for autonomous driving. CB Insights, a venture capital database, identified 30 corporate groups involved in the driverless car space, ranging from traditional automotive companies to technology brands.

In an email exchange with EE Times Asia, Edoardo Merli, Head of Greater China & South Asia and Korea Automotive Marketing and Applications at STMicroelectronics said, "There are four levels of assisted driving vehicles from basic driver assistance to full autonomous driving. In the next few years, people will see an increasing number of basic level autonomous driving cars, while the full autonomous vehicles (L4) will take a bit longer to be adopted.

"The market is moving fast along this direction, new actors (Google, Baidu, Alibaba, Apple) either alone or partnering with car makers are working on it; autonomous driving will likely be deployed initially in controlled areas (i.e. parking lots, reserved lanes for public transportation vehicles) before full adoption."

Given the attention assisted driving received at the Computex Taipei this year from both chip companies, like Nivdia, and systems integrators, like Acer, Merli's expectations may be realised sooner than later. Nvidia, in particular, was demonstrating its graphics processors were not for gaming but VR and autonomous cars.

Next: Too late in the game? >>