There were two big product announcements recently for designing autonomous automotive/car driving systems, both from the leading semiconductor companies. Below are the details of both:
1. STMicroelectronics gets into the hot area of self driving vehicle electronics on its own without acquiring any other semiconductor or other electronics technology expert, but by smartly partnering with complementing technology companies.
STMicroelectronics has developed EyeQ series of SOC chip exclusively for self driving cars. To go further advancement, Company is developing system on chip exclusively for autonomous vehicle in partnership with vision processor tech expert company called Mobileye. The new SOC they are developing is like a computer in a chip for self driving vehicles. EyeQ 5 to act as the central computer performing sensor fusion for Fully Autonomous Driving (FAD) vehicles starting in 2020.
The SOC chip is named as EyeQ5 is expected to be sampled in first half of 2018. EyeQ5 its planned to be fabricated using semiconductor technology node of 10nm or below FinFET process. EyeQ5 to have eight multithreaded CPU cores coupled with eighteen cores of Mobileye's vision processors. EyeQ5 to perform 8x times better then the current 4th generation EyeQ4. The EyeQ5 to handle more than 12 Tera operations per second with in a power consumption of 5W.
EyeQ5 can handle around 20 high-resolution sensors and for increased functional safety.
Following the industry trend of having exclusive accelerator processor cores for handling media or media stream like data, EyeQ5’s features four types of heterogeneous, fully programmable proprietary accelerator cores are optimized for computer-vision, signal-processing, and machine-learning tasks, including deep neural networks.
ST says EyeQ5 provides "super-computer" capabilities within a low-power envelope to enable price-efficient passive cooling.
On the functional safety front, EyeQ5 meets ASIL B(D) according to the ISO 26262 standard.
Featuring integrated hardware security built by Mobileye supports EyeQ5 over-the-air software updates, secure in-vehicle communication, etc.
To support the handling of continuous stream of signals from high-resolution cameras, radars, and LiDARs EyeQ5’s dedicated IOs support at least 40Gbps data bandwidth.
Its also possible to have inter-processor communication using EyeQ5 by having two PCIe Gen4 ports which could enable system expansion with multiple EyeQ5 devices or for connectivity with an application processor.
Chip also supports high-speed memory of four 32-bit LPDDR4 channels, operating at 4267MT/s.
2. 90,000 DMIPS processing capable platform for self driving cars: A highly integrated multicore processor with eight 64-bit ARM Cortex-A72 cores running at 2 GHz along with some more processor chips together offering a processing capability of 90,000 DMIPS called BlueBox engine powers in new platform for self driving cars from NXP Semiconductors.
The autonomous vehicles platform incorporates the BlueBox central computing engine, together with radar, lidar, and vision sensing, and also onboard secure V2X system.
The so-called BlueBox engine processes multiple streams of sensor data and create a complete 360° world model around the vehicle. BlueBox and its other components also provide embedded intelligence and machine learning required for complete situational assessments, supporting advanced classification tasks, object detection, localization, mapping and vehicle driving decisions.
NXP Says "unlike closed systems focused only on vision or other single-sensor data streams, the NXP BlueBox engine for autonomous vehicles is an open-platform, Linux-based solution programmable in linear C language that automotive manufacturers can easily customize to their needs for optimal product differentiation."
BlueBox delivers 90,000 DMIPS (million instructions per second) of performance at under 40 Watts of power. Processor elements include graphics engines, dedicated high performing image processing accelerators, ARM cores and APEX image processing.