Renesas lays out connected car future
The overall automotive electronics scenario is determined by the Acronym CASE – the cars are increasingly Connected, driving is Automated with Shared usage models, and with Electric propulsion, enumerated Shinichi Yoshioka, CTO of Renesas’ Automotive Business Division: 90 % of all new cars will be connected by 2025. Automated driving at levels 2 and 2.9 are going to volume production. Car Sharing and new usage models bring Big Data services into the equation. Battery, Plug-in hybrid and 48V vehicles are seeing quickly rising demand. So far, nothing new in the scenario. Yoshioka added another E to this acronym: The E/E architecture undergoes a dramatic change with centralized controls and virtualization being the dominating determinants.
This constellation highlights the requirements for future in-car compute platforms. Powerful edge computing, Artificial Intelligence and computer vision will need more sophisticated and advanced silicon. The same holds true for the link between vehicle and cloud which runs through a similarly powerful head unit or gateway computer. “These are the areas we want to be active”, Yoshioka said.
in which Renesas plans to intensify its activities
Having added the R-Car V3M and R-Car V3H processor variants to the R-Car family in the 2017 / 2018 time frame, the company is offering two devices with focus on vision applications. According to Yoshioka, these two variants will be seen in vehicles whose production starts in 2019.
The next platform variant will be R-Car-Next which is expected to populate cars with SoP (Start of Production) in 2022. The engineering parameters for the R-Car-Next make clear who Renesas regards as the main competitor: In terms of power consumption and AI performance, the R-Car-Next beats Nvidia’s Xavier platform. Indeed it is an open secret in the industry that E/E developers at automotive OEMs and tier ones worry about the power needs of Nvidia’s computing platforms.
such as Mobileye and Nvidia with more computing power
and at the same time lower power consumption. (C) for all images: Renesas
Having the silicon is one thing. The SoCs and processors are brought to life by a neural network conversion tool that supports the entire workflow, starting with an image data base for training; generating neural networks in the cloud and generating application code through to finished AI applications for the Renesas V3X processor family.
Yoshioka also introduced Renesas’ approach to enable robust systems for automated driving (AD). The concept, called RIDs (Renesas Interconnected and Databank Solution), enables synchronization of vital AD functions by providing a unified time-base to attach a timestamp to critical data. It also enables the automotive in-car electronic system to switch communication route automatically – thus it can support fail-operational conditions.
For the future, Renesas plans to further strengthen the computing power of its R-Car platform. Between 2020 and 2022, various variants of the Next Gen R-Car Vx processors are to be launched, covering the market for ADAS and autonomous vehicles. Features of this generation are the use of the latest CPU cores Enyo from Arm (up to eight cores) as well as a powerful AI engine. Designed as a Convolutional Neuronal Network, this engine will deliver up to 60 Tera Ops. And yes, the Next Gen R-Car Vx will be upward compatible from the current V3M / V3H versions.
Related articles:
Renesas IP licensing eyes AI, autonomous markets
Software eases virtualization for Renesas’ R-Car platform
NXP, Kalray defy Nvidia with automotive AI computing platform
Daimler takes Nvidia inside car central computer
Nvidia Drive AGX platform on-board autonomous vehicle fleet
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
