ARM: Announced new products from images for in -vehicle image signal processors to expand the hiring of driving and automation technology

By combining with the "Cortex-A78AE" CPU and "Mali-G78AE" GPU, Mali-C78AE realizes a complete vision pipeline of the advanced operation support system (ADAS), optimizes performance and minimizes power consumption.Provide a consistent approach to functional safety while suppressing.As a result, we will support the ADAS mass market deployment to the next stage.

　According to a recently published report by Strategy Analytics, the value of the in -vehicle camera market is expected to grow more than 19%from 2020 to 2025, and the data is necessary for cars to judge the peripheral situation.It is the most important type of sensor above.

　ARM's Automotive / IoT Division, Automotive Vice President, said: CHET BABLA:"As the number of in -vehicle cameras is increasing and the altitude is increasing, it is higher for efficient and secure image data generated to obtain a variety of requirements of machine vision and human vision.It is required to have an arithmetic ability. Mali-C78AE responds to the needs of the industry for new approaches for image processing and realizes new features in ADAS and autonomous driving. "

■ In both human vision and machine vision, safety is a top priority Mali-C78AE, which is specially designed for both human vision and machine vision required for many ADAS functions, up to 4 units.It is possible to process data from real -time cameras or 16 virtual cameras.

　Mali-C78AE is developed from zero in the form of a hardware safety mechanism and diagnostic software function, which allows system designers to support the "ISO 26262 ASIL B" functional safety requirements.Mali-C78AE attempts to prevent and detect the failure in a single camera frame, which can provide the incorrect processing of the frame data.For this reason, Mali-C78AE adopts 380 or more failure detection circuits, supports continuous embedded self-tests, and can detect the connected camera sensor and hardware failure.

■ Visual information has a large amount of data, and the workload is a high load, as well as safety and user experience, and the processing speed is also an important factor in Mali-C78AE.The process of acquiring an image with the sensor, processing from ISP to GPU, and displaying the display for drivers must be completed in 150 milliseconds.Beyond this, the driver recognizes the delay in the case of a parking assist.In the case of machine vision applications, the allowable travel distance is 250mm from acquiring camera images to decision -making processing.Beyond this, the machine vision system is too slow and it is impossible to respond to the driving situation that requires accurate and timely judgment.

　The ADAS camera needs to collect the most relevant information from each frame so that drivers and machines can make the best decisions.Mali-C78AE adopts advanced noise reduction technology and dynamic range management to ensure clear display and appropriate exposure of each frame by adjusting excessively dark and bright areas in the frame.Mali-C78AE enhances the requirements for memory, communication, and processing by performing real-time processing of camera data through up to four high-resolution and high-frame rates.Realize.

　Machine vision applications such as lane deviation alerts are not supported by images for human visions such as surround views, so set the camera individually to implement multiple ADAS functions at this time.There is a need.In order to reduce the implementation cost of these multiple ADAS functions, Mali-C78AE performs a downscale and color conversion of the output of the sensor optimized for the machine vision to generate images that match the human eye.Make the camera sensor correspond to both purposes.OEM can save costs and reduce complexity by eliminating the duplication of the camera and the accompanying electronic circuits and wiring.By introducing a wide range of camera -based ADAS functions for a wide variety of vehicles, a safer and better driving experience can be achieved.

■ Pioneer companies in the industry adopt powerful vision pipelines for ADAS Mobileye, a pioneer in vision safety technology for in-vehicles, for the company's next-generation EYEQ technology, the new Mali-C78AE ISPImplemented.When Mobileye started developing Mobileye Eyeq Ultra and EYEQ6H, Mali-C78AE was adopted to efficiently process image data.By combining this product with the Mali-G78AE GPU, a real-time intuitive graphics and rendering function that is necessary to meet the strict requirements of these applications will be realized.

　ELCHANAN RUSHINEK, an executive vice president in Mobileye's engineering, said:"Our surround view visualization function that uses ARM's ISP and GPU technology combinations enables centralized operation -type operations to provide state -of -the -art ADAS and parking assist systems with a single SoC."