NXP S32V Processor for Vision and Machine Learning Applications Now Available

S32V processor enhanced with Vision SDK and expanded ecosystem enablement partners

• Offers powerful GPU for surround view image stitching and offloads advanced vision processing and machine learning algorithms to APEX-2 engines
• Robust sensor fusion solution with multiple layers of functional safety and security
• High-performance per power envelope (GHz/mW) for low power ADAS vision processing, for adding additional capabilities within the same system environment

AUSTIN, TEXAS Sept.13, 2017 - NXP Semiconductors (NASDAQ:NXPI) today announced the full availability of the NXP S32V processor along with complementary S32 Design Studio IDE enablement to simplify and accelerate the deployment of ADAS vision and neural network solutions in automotive, transportation and industrial applications.

Success in the race to fully autonomous level-5 vehicles will depend on how accurately a car can interpret its surroundings with the fusion of vision and sensor data. Systems such as the NCAP front camera and surround view, which the S32V processor is designed for, will also have the ability to locate, identify, track and classify multiple objects simultaneously.

Neural networks and machine learning are critical to develop autonomous vehicles since they will need to adapt to changing environments and adverse conditions. These technologies are being deployed into multiple industries to identify speech, recognize objects, understand behavior and predict events. As processing scales to meet the needs of autonomous driving, designers will need architectural flexibility to add differentiation and provide clear safety certification that S32V processors provide.



The S32V Processor

• Meets the needs of autonomous vehicle development with its blend of safety fusion and high performance processing using NXP's second generation APEX-2 vision accelerators.
• Includes dual APEX-2 engines with 128 parallel computational vector units that process image data faster and at lower power than traditional GPU based solutions.
• APEX-2 engines are easily programmed using APEX-CV library for computer vision algorithms.
• Included in the Vision SDK are APEX-CV base level and APEX-CV pro library algorithms as Canny, ORB, Harris, HOG, Gaussian, Hough and other detection algorithms to simplify development.

Quote

"The launch of the S32V processor provides a dedicated solution for vision processing and machine learning to a wide range of automotive and industrial customers," said Paul Lee, global distribution marketing manager for NXP's Automotive Microcontroller and Processors Business Line. "We look forward to the next generation of innovative applications that design engineers will develop with our Vision SDK and software enablement tools."

NXP S32V Vision Processor Specs

• Low power – <10-watt vision processing solution offers customers a passive cooled MPU with or without GPU enabled
• Performance – offers up to 10k DMIPS with Quad 1GHz A53 cores and additional performance from GC3000 Vivante GPU and dual APEX-2 vision accelerators
• Functional safety for sensor fusion – M4 MCU core, safety clusters, end-to-end ECC, FCCU and FMEDA included to meet ISO26262 and IEC61508
• On chip Image Sensor Processor (ISP) provides opportunity to save 30 percent in BOM costs by using cheaper camera sensors to capture image data
• Security – Cryptographic Service Engine (CSE) meets the SHE protocol specification and 128-bit encryption and true RNG

S32V Development Tools

• Complimentary compiler and debugger S32 Design Studio IDE for Vision v2.0
• Software development kit Vision SDK v1.0 with integrated APEX and ISP graphical design tools
• Board support package Linux BSP for both low-cost evaluation boards (SBC-S32V234 and S32V234-EVB2)
• Project code Demo code examples are included in VSDK such as lane departure warning (LDW), Optical Flow, Stereo Vision, Squeezenet™ neural networks, Face Detection and more
• Operating system support options Linux (Yocto), SafeRTOS and AutoSAR OS
• Partner network for training and development support

For More Information on the NXP S32V

Visit www.nxp.com/s32V to learn more about our S32V vision processor, SBC-S32V234 development platform and the complimentary S32 Design Studio IDE to compile and debug with integrated Vision Software Development Kit and code examples for lane departure warning, pedestrian detection and convolutional neural network for machine learning and object classification.

S32V vision processor samples are available today.

Video

S32V Launch Video