Renesas targets high volume vision AI with DRP AI core

eenewseurope.com, Mar. 13, 2025 – 

Renesas Electronics has launched an integrated quad core ARM microprocessor with its proprietary DRP embedded AI technology for high volume vision applications.

The RZ/V2N uses the Dynamically Reconfigurable Processor DRP-AI3 with 10 TOPS/W and AI inference performance of up to 15 TOPS through its advanced pruning technology.

The device is significantly smaller than the high end 80TOPS RZ/V2H, with its package area measuring only 15 mm square, reducing the area required for mounting by 38%. THe higher energy efficiency avoids the need for a additional cooling fans and reducing the size and cost of embedded systems for vision AI in a wide range of applications, from AI cameras for traffic and congestion analysis in commercial facilities, industrial cameras for visual inspection on production lines, to driver monitoring systems for behaviour analysis.

Renesas is showing a demonstration of the RZ/V2N at embedded world 2025 in Nuremberg, Germany, this week.

The RZ/V2N has four ARM Cortex-A55 CPU cores and a single Cortex-M33 core, coupled with a high-quality image signal processor (ISP), Arm Mali-C55. The RZ/V2N also has two channels of MIPI camera interfaces, which allow two cameras to be connected to capture double-angle images.

A two-camera system significantly improves spatial recognition performance compared to a single camera system and enables more precise human motion line analysis and fall detection. This can also capture images from different locations, allowing a single chip to count cars in a parking lot and recognize license plates efficiently.

 “Since launching the RZ/V2H last year to target next-generation robotics requiring vision AI and real-time control, we have received tremendous interest in using the Renesas DRP-AI accelerator,” said Daryl Khoo, VP of Embedded Processing at Renesas.  “With the addition of the RZ/V2N, sharing the same lineage as the RZ/V2H, we are extending our reach to address mid-range applications, particularly endpoint vision AI that does not need to be realized with power hungry designs.”

Click here to read more...

 Back