It's called RH850 / V1R, and its goals are driver assistance and autonomous driving. "The radar sensor requires ADAS (driver assistance) applications, including emergency braking and adaptive cruise control, because the radar sensor is not adversely affected by severe weather conditions, such as rain, fog or direct sunlight," Reza said. In autonomous driving, "accurate sensing is essential to meet the requirements of distance resolution, object separation, and measurement speed. This requires increasing the number of antennas to improve signal processing performance."
The first is the RH850 / V1R-M, which is used for medium and long-range radars, including DSPs, to process raw data to the target object. It will run radar-related algorithms such as FFT, beamforming, windowing, channel calibration and peak search. It then created an automotive radar math library for algorithm developers.
The CPU is designed to handle security-related classification and tracking. There are two G3MH CPU cores running at 320MHz, and 2Mbyte of 40nm embedded flash memory and 2Mbyte RAM. The company states: "By integrating 2M bytes of RAM, it can process radar cube data such as distance and speed FFT, digital beamforming," constant false alarm rate "and peak detection. The highest performing core in the RH850 series is superscalar RISC The architecture has two 7-level integer pipelines, allowing two different instructions to be executed simultaneously, each G3MH core reaching 3.2DMIPS / MHz.
IC samples and DSP math libraries will be available in the second half of 2017. Mass production is scheduled to begin in November 2018. A C / C ++ compiler, debugger, simulation model and performance analysis tools will also be provided. -But availability is subject to change without notice.
