Renesas Electronics Corporation introduced an update of its R-Car V3H system-on-chip (SoC) to deliver significantly improved deep learning performance for smart camera applications, including driver and occupant monitoring systems (DMS / OMS); automotive front cameras; surround view; and auto parking for high-volume vehicles up to Level 2+.
The updated SoC combines sensor fusion on the real-time domain with up to ASIL C metrics and an architecture optimized for smart computer vision. It offers OEMs and Tier 1s a high-performance, low-power solution that supports the latest NCAP 2020 requirements as well as the roadmap to NCAP 2025 3 Stars at competitive system costs.
Building on the state-of-the-art recognition technology introduced with the R-Car V3H in February 2018, which includes integrated IP for convolutional neural networks (CNN), the updated R-Car V3H delivers four times the performance for CNN processing compared to the earlier version and achieves up to overall 7.2 TOPS processing including all computer vision IPs while maintaining low power consumption levels.
A CNN is a deep learning algorithm which can take in an input image, assign importance (learnable weights and biases) to various aspects/objects in the image and be able to differentiate one from the other.
The highly integrated SoC supports up to ASIL C metrics safety goals on the real-time domain, reducing the need for an external safety microcontroller (MCU) to manage sensor fusion and final decision actions.
R-Car V3H computer vision pipeline
The R-Car V3H features a suite of proven IPs that support the perception stack, sensor fusion with radar and/or Lidar, and ISP with up to eight MP cameras support, allowing OEMs and Tier 1s to achieve quicker time to market with lower bill of materials costs.
Key Features of R-Car V3H SoC:
Delivers up to 7.2 TOPS, including 3.7 TOPS for CNN, with optimized performance-to-power balance;
Supports ASIL D development process for systematic capability for the full SoC;
Supports metric targets for ASIL B (sensor layer, application processors) and ASIL C (real-time domain) safety goals;
Full hardware and software compatibility with the currently mass-produced R-Car V3H SoC;
Integrates a full set of automotive peripherals including CAN, Ethernet AVB, and FlexRay; and
Features a full set of video processing and image recognition IP for advanced sensing and recognition, including CNN-IP, Computer vision engine, Image-distortion-correction IPs, Stereovision, Classifier and Dense Optical Flow.
The updated R-Car V3H joins Renesas’ expanding lineup of SoCs within the open and flexible Renesas autonomy platform, which offers OEMs and Tier 1s complete scalability from entry-level NCAP applications up to highly automated driving systems. Customers can also combine the R-Car V3H SoC with Renesas’ high-performance low-power RH850 MCU, integrated power management ICs, and power transistor devices to access the key components required for more complex safety compliant systems.
Mass production of the updated R-Car V3H is planned for first quarter of 2022.