Machine Learning and Sensor Fusion Demonstration at CAR-ELE Japan 2017
With Zynq SoC and Zynq UltraScale+ MPSoC devices

Xilinx Korea is showcasing advanced driver assistance systems (ADAS) and autonomous driving (AD) solutions at the International Automotive Electronics Expo (CAR-ELE Japan) being held in Tokyo.

Xilinx also demonstrates machine learning, computer vision, and sensor fusion on Zynq SoC and Zynq UltraScale MPSoC devices.



The demos to be showcased by Xilinx and ecosystem partners include: deep learning using CNN on Zynq UltraScale MPSoC; a camera-based driver monitoring system; a multi-camera system using Ethernet audio/video bridging; an advanced camera and display-based E-Mirror system; and a high-end surround view system.

We demonstrate the use of a Zynq UltraScale MPSoC with deep learning as an embedded computing platform to detect pedestrians and show real-time machine learning tracking the head and eyelids to accurately measure the direction of gaze.

In addition, Ethernet-AVB for automotive camera systems where multiple camera modules based on the Zynq-7000 All Programmable SoC are used for image signal processing, object detection, distortion correction, etc. We will demonstrate sensor fusion for an advanced surround view virtual flight camera system that enables the driver to seamlessly transition to a 3D view of the vehicle.

Finally, we introduce a method to replace the standard rearview mirror of a car or other vehicle and demonstrate how input information from multiple cameras is combined with the display.