The joint Hailo-Renesas solution is intended to make sophisticated ADAS technology more accessible in cars of all types.
The collaboration sees Hailo-8 processors, designed for heavy Neural Network (NN) acceleration, combined with the Renesas R-Car V3H & R-Car V4H systems-on-chip (SoCs), to deliver robust, scalable, efficient, and cost-effective ADAS solutions for zonal and centralised vehicular ECUs.
The solutions are scalable for a wide range of ADAS functions from ADAS L2+ up to L4 AD functions requiring very high tera-operations per second (TOPS), while delivering best-in-class TOPS/W.
“Technological challenges have caused great difficulties for automakers looking to make advanced functions accessible in anything but higher-end vehicles,” said Orr Danon, CEO and Co-Founder of Hailo. “Together with Renesas, we are offering a solution helping democratise ADAS, so its benefits can be enjoyed by more and more drivers. AI will play a major role in making our roads safer.”
“Automotive OEMs are looking for scalable solutions for ADAS and AD to streamline software development,” added Yusuke Kawasaki, Director, Automotive Digital Products Marketing Division at Renesas. “We are confident that our collaboration with Hailo, which has expertise in AI technology, will enable our customers to easily extend the performance of the R-Car so that they can flexibly and quickly incorporate ADAS/AD functionality in various classes of vehicles.”
Key features for the joint Hailo-Renesas solution include:
- Independent scalability in AI & compute, allowing flexibility and cost-effectiveness for Level 2+ to Level 4 ADAS designs
- The highest performing solution available today in the automotive market, ranging from 60 TOPS to over 170 TOPS for multi-sensor perception and fusion
- Best-in-class power efficiency, enabling ECUs to be passively cooled, which reduces the BOM and system cost
- Functional safety for the ADAS/AD ECU with Renesas R-Car (R-Car V4H is expected to achieve ASIL-B and D metrics for the real-time domain) and Hailo (Hailo-8 supports ASIL B)
- An open software ecosystem allowing OEMs/Tier-1s more control & possibility for innovation
Featuring up to 26 TOPS at a typical power consumption of 2.5 W, the Hailo-8 chip is built with an architecture that enables edge devices to run sophisticated Deep Learning applications that could previously only run on the cloud.
The joint Hailo-Renesas solution is available for customer evaluations and demonstrations.
