Renesas and StradVision Team Up for Machine Learning Smart Cameras in ADAS Applications
Renesas and StradVision are leveling up automotive camera solutions with deep learning for object recognition on low-power SoCs.
Renesas Electronics Corporation and StradVision have announced the joint development of a deep learning-based object recognition solution for smart cameras.
The venture is aimed at next-generation advanced driver assistance system (ADAS) applications and cameras for ADAS Level 2 and above. ADAS implementations require high-precision object recognition to detect vulnerable pedestrians and cyclists and must do so while consuming very low power.
Image from Renesas
Through the combination of StradVision’s ADAS implementation experience and Renesas’ R-Car SoCs, both ends are achieved, which will serve to facilitate the widespread adoption of ADAS.
The Hardware: Renesas R-Car SoCs
Renesas "R-Car" is the company’s lineup of system-on-chip (SoCs) for car information systems.
The roadmap Renesas has been following to keep up with the rapid evolution of automotive technology. Image from Renesas
The latest versions of the chips include:
- R-Car H3 is an automotive computing platform for autonomous driving. It provides cognitive computing capabilities and can keep up with the large volumes of information from vehicle sensors accurately an in real-time. This powerful SOC runs the applications enabling obstacle detection, driver status recognition, hazard prediction, and hazard avoidance.
- R-Car V3M, an SoC that is primarily for front camera applications, and also surround view systems or LIDARs (light detection and ranging – a sensing method based on pulsed lasers).
- R-Car V3H performs the simultaneous recognition of vehicles, people and driving lanes. It can process image data at a rate of 25 frames per second, enabling swift evaluation and POC development.
The block diagram for the R-Car H3 SiP module. Image from Renesas
In addition to its new emphasis on automotive intelligence, Renesas is involved in solutions for building automation, energy management systems, healthcare, home appliances, industrial automation, and office automation. The company is also a source for semiconductor and development environments for a variety of disciplines, including MCUs, sensors, power management, optoelectronics, aerospace, and harsh environments.
The Software: StradVision Machine Learning Algorithms
StradVision is a Korea-based company with offices in Seoul, Tokyo, and San Jose, CA.
In addition to working with Renesas SoCs, StradVision’s V3SVNet has also been ported onto over fourteen platforms, including TI’s TDA2X ADAS SoC. The company cites that its SVNet External can execute functions including the recognition of traffic signs, pedestrians, and cyclists, as well as issuing warnings regarding impending forward collisions, headway collisions, and lane departure.
StradVision’s SVNet Internal can detect both driver and passenger's status including drowsiness, height, and weight, which in turn can be applied to adjust airbag pressure, seat position, angle, and much more. It, too, is available for TI’s TDA2/3x, Qualcomm SDM855, ARM CPUs as well as Rensas’ V3M and V3H.
Machine Learning for Assisted Driving
2019 has been an exciting year for assisted driving, but team-ups between hardware and software companies in the service of ADAS has been ongoing for years. In 2017, for example, Nextchip and CEVA joined forces to pursue machine vision for automotive applications.
Even so, the pairing of Renesas and StradVision is a sign of continued interest in developing automated driver assistance for the next generation of vehicles.
“A leader in vision processing technology, StradVision has abundant experience developing ADAS implementations using Renesas’ R-Car SoCs, and with this collaboration, we are enabling production-ready solutions that enable safe and accurate mobility in the future,” said Naoki Yoshida, Vice President of Renesas’ Automotive Technical Customer Engagement Business Division.
Junhwan Kim, CEO of StradVision, is equally sanguine. “This joint effort will not only translate into quick and effective evaluations, but also deliver greatly improved ADAS performance."
Kim anticipates "massive growth expected in the front camera market in the coming years"—a good bet by any measure.
Have you worked on autonomous vehicle projects? How has machine vision changed your job? Tell us about your experiences in the comments below.