Arm has unveiled a new automotive ISP to meet industry requirements for a new approach to image processing.

Arm announced the launch of the Mali-C78AE ISP (image signal processor) on the 21st.

The Mali-C78AE ISP has been added to an IP portfolio specifically developed to meet the performance and safety standards of automotive applications.

This provides a complete Advanced Driver Assistance System (ADAS) vision pipeline to optimize performance, minimize power consumption, and provide a consistent approach to functional safety.

According to a recent report by Strategy Analytics, the value of the automotive camera market is projected to grow by more than 19% from 2020 to 2025.

The Mali-C78AE is expected to become the most important type of sensor for providing the data necessary for vehicles to make decisions about their surrounding environment.

Chet Babla, Vice President of Automotive at Arm’s Automotive and IoT Business Unit, stated, “As the number and sophistication of automotive cameras increase, so does the computing power required to efficiently and safely convert high-throughput image data to meet the diverse requirements of machine and human vision.” He added, “The Mali-C78AE meets the industry’s demand for new approaches to image processing and will demonstrate new capabilities in the fields of ADAS and autonomous driving.” He said, "thing."

■ Human and Machine Vision, Safety is the Top Priority

The Mali-C78AE is designed to handle both human and machine vision safety applications required for various ADAS functions and can process data from up to 4 real-time cameras or 16 virtual cameras.

Mali-C78AE was developed from the outset with hardware safety mechanisms and diagnostic software functions to enable system designers to meet ISO 26262 ASIL B functional safety requirements.

In addition, it aims to prevent or detect defects in a single camera frame that may cause frame data to be misprocessed.

To this end, the ISP is equipped with over 380 fault detection circuits and a continuous self-test function, and can detect sensor and hardware defects in connected cameras.

■ Vision of Rich Data and Workloads

Just as important as safety and user experience is processing speed, a key element of the Mali-C78AE.

The process of an image collected from a sensor passing through an ISP and then through a GPU to be displayed on the driver's screen must be completed within 150ms.

If it takes longer than this, the driver may feel inconvenienced when using features such as parking assist.

In addition, the vehicle must not move more than 250 mm until the camera images collected from the machine vision application are transmitted for decision-making.

If it moves further than this, it means the machine vision system is too slow to react in driving situations where accurate and timely decisions are critical.

To enable the driver and the machine to make the best decisions, the ADAS camera [monitors] each frameYou must collect the most relevant information possible from.

The Mali-C78AE adopts advanced noise reduction technology and dynamic range (DR) management technology to adjust areas that are excessively dark or bright in the frame, ensuring that each frame is sharp and correctly exposed.

The Mali-C78AE can process camera data in real time from up to four high-resolution and high-frame-rate cameras, enabling the construction of a more efficient system.

Today, individual camera settings are required to implement various ADAS functions.

This is because cameras used in machine vision applications, such as lane departure warnings, do not generate images suitable for human vision like surround view.

To reduce the cost of implementing various ADAS functions, the Mali-C78AE enables the camera sensor to be used for dual purposes by downscaling the output of machine vision-optimized sensors and converting colors to generate images suitable for the human eye.

In addition, OEMs can reduce costs and complexity by avoiding redundant use of cameras, related electronics, and wiring.

As a result, camera-based ADAS features can be deployed more widely across various car models, providing drivers with a safer and enhanced user experience.

■ Powerful Vision Pipeline for ADAS

Mobileye, a leader in automotive vision-safety technology, is taking the lead in implementing the new Mali-C78AE ISP into Mobileye's next-generation EyeQ technology.

Image of when Mobileye started developing Mobileye EyeQ Ultra and EyeQ6H We adopted Mali-C78AE to process data efficiently.

In addition, by using the Mali-G78AE GPU, it has enabled the safety features and smooth, intuitive real-time graphics rendering capabilities necessary to meet the demanding requirements of applications.

Elkanan Rushnek, Vice President of Engineering at Mobileye, stated, “By combining Arm’s ISP and GPU technologies for surround view visualization capabilities, Mobileye is now able to provide a centralized computing solution that delivers cutting-edge ADAS and parking assistance systems on a single SoC.”