Arduino Projects   |   Raspberry Pi   |   Electronic Circuits   |   AVR   |   PIC   |   8051   |   Electronic Projects

Renesas Electronics Develops Embedded Automotive ADAS System with Perception Quick Start Software

Submitted By: 

Shreepanjali Mod

The leading supplier of highly developed semiconductor solutions, Renesas, is sprinting development of its ADAS (Advanced Driver Assistance Systems) with the newly developed Perception Quick Start Software. This new software is rooted in R-Car V3H SoC (System-on-Chip). This new solution will be delivering reference software for COD (Camera Obstacle Detection), RFD (Road Feature Detection), LiDAR Obstacle Detection (LOD). These are the three areas of recognition for sensor-based Level 2+ autonomous vehicle systems. 

The R-Car V3H SoCs are able to yield a unique combination of artificial intelligence processing and computer vision performance while consuming much less power offering a balanced embedded solutions for automotive front cameras in Level 2+ autonomous vehicles. Renesas designed the SoCs with highly dedicated hardware accelerators for most crucial algorithms including the object classification, stereo disparity, dense optical flow, and neural networks. 
This new perception software offers a complete pipeline reference for developers working on complicated but pocket friendly and power-efficient accelerators pacing way out for starting application design for experts as well as beginners and amateurs. The reference software takes care of input from recorded data or sensor data and all processing stages. 
Key Features
The LOD software makes use of CV-E and CNN-IP in detection of 3D objects including trucks and cars. The LOD can reach up to 15 FPS with 3D bounding boxes for 50 meters
The COD reference software makes use of CNN (Convolutional Neural Network) IP, IMR (Image Rendering) technology, and CV-E (Computer Vision Engine) for 2D object detection like for pedestrians, trucks, buses, and cars. It can reach upto approx. 30 FPS (Frames Per Second). 
Fig. 1: Image Showing Example of COD
The RFD reference software also makes use of IMR, CV-E, and CNN-IP as well as a versatile pipeline engine for identification of drivable free space, distances to lanes, road boundaries, and lanes to back NCAP 2020. It can achieve up to 30 FPS. 
In Maker’s Words
The director of Automotive Advanced Systems Innovation Department of Renesas, Tim Grai, says, “Specialized hardware accelerators play an essential role in achieving the computer vision performance and accuracy required in embedded ADAS and autonomy applications while still meeting stringent in-vehicle power consumption limits: however, the complexity of these accelerators can present a steep learning curve. With the new Perception Quick Start Software, we are able to offer a set of application software along with the underlying primitives to simplify the use of these complex accelerators needed to achieve embedded ADAS.”