Engineering 101

Accelerating Intelligent Computer Vision Processing SoCs

0

A collaboration has been announced between Synopsys, and Kudan to optimize Kudan’s SLAM computer software algorithms for Synopsys’ DesignWare ARC EV6x Embedded Vision Processor IP. Kudan’s KudanSLAM software algorithms create and expand a map of an environment while simultaneously keeping track of the camera’s location within it.

SLAM is used in autonomous vehicles, drones, robots, and augmented reality applications for concurrent scene mapping and device localization.

Kudan is optimizing its software to offload and accelerate algorithms by taking advantage of the Synopsys DesignWare ARC EV6x Embedded Vision Processors’ scalable hardware architecture, which includes up to four 512-bit vector DSPs and a fully programmable convolutional neural network (CNN) engine.

The combined hardware-software solution enables designers to accelerate the SLAM tasks of tracking and mapping that take input from LiDAR, Time of Flight (TOF) cameras, inertial measurement units (IMUs), or odometry data while consuming significantly less power and memory resources than alternative implementations.

“Solutions that increase the efficiency and accuracy of computer vision devices will form the foundation of mass-market SLAM applications,” said Tomo Ohno, Founder and Managing Director at Kudan. “The combination of the ARC EV6x Embedded Vision Processor and KudanSLAM will deliver the best performance and power efficiencies in the market and provide the software infrastructures by which product builders can develop revolutionary applications for automotive ADAS, robotics, and AR/VR.”

The fully programmable and configurable DesignWare ARC EV6x Embedded Vision Processor IP cores are optimized for embedded vision applications, combining the flexibility of software solutions with the low cost and low power consumption of hardware. The ARC EV6x Embedded Vision Processor family is supported by Synopsys’ MetaWare EV Development Toolkit, a comprehensive software programming environment based on common embedded vision standards including OpenCV, OpenVX, and OpenCL C.

“Performance, power consumption, and high accuracy are critical requirements for computer vision applications,” said John Koeter, Vice President of Marketing for IP at Synopsys. “By collaborating with Kudan to optimize their software for Synopsys’ ARC EV6x Embedded Vision Processor, we are providing designers with a highly efficient and accurate hardware-software solution to accelerate the development of SoCs implementing SLAM functionality.”

The DesignWare EV6x Embedded Vision Processors are available now. KudanSLAM software optimized for EV6x Vision Processors is planned to be available in the second half of 2019.

Leave A Reply

Your email address will not be published.