Ambarella’s next-gen AI SoCs for fleet dash cams and vehicle gateways enable vision language models and transformer networks without fan cooling

Ambarella an edge AI semiconductor company, announced during AutoSens USA, the latest generation of its AI systems-on-chip (SoCs) for in-vehicle fleet telematics systems. Ideal for AI dash cams integrating front-facing ADAS and driver monitoring systems (DMS), the new CV75AX provides up to 2x the AI performance over Ambarella’s prior-generation SoC, enabling the latest transformer neural networks for improved accuracy and reduced false positives without having to train for every object. The CV75AX also provides the AI horsepower for more complex driver-scoring models. Well suited for next-gen vehicle gateways, the new CV72AX supports up to 10 cameras and offers up to 6x the AI performance over its predecessor, enabling both vision transformers and multi-modal vision language models (VLMs), which provide instant event alerts, as well as multi-channel video pre-categorization and natural-language searches for more efficient video analysis—all without the need for training.

“Our current generation of AI SoCs has been widely adopted for in-vehicle systems by fleet telematics providers, with our unique ability to enable fanless AI dash cams that don’t impede the driver’s view,” said Senya Pertsel, sr. director of automotive marketing at Ambarella. “With our new CV75AX and CV72AX SoCs’ increased AI performance, efficiency and features, while maintaining that fanless capability, we have already secured designs and are being evaluated by telematics providers for next-generation systems to enable transformer-perception, VLM-analysis and driver-scoring models.”

Both of these new fleet telematics SoCs feature Ambarella’s latest CVflow® 3.0 AI engine, 2x the CPU performance, LPDDR5 support with 2x the DRAM bandwidth and a higher-speed USB 3.2 interface for 5G modems, compared to the prior generation. They also integrate the latest generation of Ambarella’s image signal processor and H.264/5 video encoder, providing high image quality for both human viewing and computer vision, as well as multi-channel recording for evidence collection. Additionally, these are the latest Ambarella SoCs to be manufactured in Samsung Foundry’s 5nm FinFET process technology, enabling them to operate over a wide temperature range (-40C to 105C ambient), which is critical for automotive applications; and further contributing to their lower power consumption and increased performance compared to the prior generation.

With their increased AI performance, these new SoCs are also bringing Ambarella’s advanced AI-based high dynamic range (HDR) image processing technology to in-vehicle fleet telematics systems. Known as Vivid HDR, this proprietary technology uses AI to augment and improve the image quality in scenes with widely contrasting light and dark areas.

The CV75AX’s on-chip microcontroller (MCU) core has a new ultra-low-power sleep mode where the SoC consumes less than 5mw. This eliminates the need for an external MCU to monitor the parked vehicle’s sensors without battery depletion, reducing BOM cost and design complexity.

To accelerate the development of in-vehicle fleet telematics applications, the CV72AX and CV75AX use the same SDK and are supported by Ambarella’s Cooper™ Developer Platform. This recently introduced platform provides comprehensive hardware and software solutions for creating edge AI systems, including powerful, safe and secure compute and software capabilities. It consists of industrial-grade hardware tools, collectively called Cooper Metal; along with Cooper Foundry, which provides a multi-layer software stack that supports Ambarella’s entire portfolio of AI SoCs.