Renesas unveils next-generation R-Car SoC for ADAS and AD

Renesas Electronics Corp. has introduced its R-Car V4H system on chip (SoC) for central processing in advanced driver-assistance (ADAS) and automated driving (AD) solutions. Targeting Level 2+ and Level 3 automated driving, the new R-Car SoC touts deep learning performance of up to 34 tera operations per second (TOPS), which enables high-speed image recognition and processing of surrounding objects by automotive cameras, radar, and LiDAR.

Click for a larger image. (Source: Renesas Electronics)

“The R-Car V4H is the next step in our long line of high performance SoC products for the ADAS/AD segments,” said Daniel Sisco, senior director automotive systems at Renesas Electronics. “We expect the R-Car V4H to extend the success we found with our R-Car Gen3 products such as V3H and V3M. The V4H will enable us to attack the latest requirements and market demands such as more powerful AI/CNN [artificial intelligence/convolutional neural network] capability and enhanced safety features.

The R-Car V4H offers improvements over the Gen3 parts, but also share common architecture points to enable software re-usability and compatibility, said Sisco. “Some major changes include upgraded AI capability, the addition of a 3D GPU for surround view and other rendering tasks, the latest in hardware accelerators such as ISP and IMR, and the latest in communication interfaces.”

The SoC is built around  four Arm Cortex-A76 cores at 1.8 GHz with a total of 49k DMIPS of general compute for ADAS/AD applications, three lockstep Arm Cortex-R52 cores at 1.4 GHz with a total of 9k DMIPS to support ASIL D real-time operation and eliminate the need for external microcontrollers, along with the dedicated deep learning and computer vision IPs. It also includes an image renderer (IMR) fisheye distortion correction or other mathematical operation, a graphic processor unit GPU (AXM-8-256 at 600 MHz, for a total of over 150 GFLOPS), two fourth generation PCIe interfaces, and several dedicated automotive interfaces, including CAN, Ethernet AVB, TSN, and FlexRay.

R-Car V4H SoC block diagram. Click for a larger image. (Source: Renesas Electronics)

Sisco said the R-Car V4H targets the highest volume applications for automated driving. “The highest volume adoption of ADAS /AD can be found in vehicles with SAE Level 2+ and SAE Level 3 applications, such as automated parking or NCAP compliance.”

He also noted that Level 3 features are in many production vehicles today with some already supported by the R-Car V3H SoC. The V3H and V3M focus on smart front cameras for Level 1 and Level 2 applications.

Renesas automated driving product roadmap. Click for a larger image. (Source: Renesas Electronics)

Thanks to a high level of integration, the R-Car V4H allows customers to develop single-chip, ADAS electric control units (ECUs). These control units may support driving systems for automated driving Levels 2+ and Level 3, including full NCAP 2025 features. The R-Car V4H also supports surround view and automatic parking functions with 3D visualization effects such as realistic rendering. The signal processing portion of the R-Car V4H is expected to achieve ASIL B and D metrics for the real time domain.

As part of the integration, Renesas provides a dedicated power solution for R-Car V4H, which is based around the RAA271041 pre-regulator and the RAA271005 power management IC (PMIC). This provides a highly reliable power supply for the R-Car V4H and peripheral memories from the 12-V supply of the vehicle battery, while enabling low power operation, targeting ASIL D compliance for systematic and random hardware faults at very low BOM cost, said Renesas. The upshot: It reduces hardware and software development efforts and design complexity.

Renesas anticipates the SoC will help enable the development of the software-defined vehicle. “The software-defined car must integrate the latest applications and features from multiple segments, including ADAS/AD,” said Sisco. “The R-Car V4H will support this by enabling functions such as NCAP compliance, surround view, automated parking, and front camera applications.”

He also noted that another critical aspect of the software-defined car is the ability to easily update software over time using over-the-air software update capabilities, which the R-Car V4H supports,

Samples of the R-Car V4H SoC are available now, with mass production scheduled for the second quarter of 2024. A R-Car V4H software development kit (SDK) is also available for initial device evaluation and software development including deep learning. The SDK offers full functionality for machine learning development, and optimization of embedded systems for performance, power efficiency, and functional safety. Complete simulation models are available.