ADAS Computing on a Greater Scale

Samsung engineers are developing next-gen domain controllers to meet specific performance needs of automated/autonomous driving.

Development engineer Matt Lewis testing the DRVLINE computing platform recently on California public roads. (Credit: Samsung)

Open. Modular. Scalable. Those words describe the framework of Samsung Electronic’s autonomous vehicle platform currently under development.

The new compute platform “is being designed as a functionally-safe ADAS (advanced driver-assistance systems) domain controller that can be scaled for the performance needs of different autonomous driving levels,” noted Dave Anderson, Director of Samsung Electronics’ Autonomous Driving Program, during a recent interview with SAE’s Autonomous Vehicle Engineering.

Called DRVLINE, the modular architecture is interconnected with multiple buses, including Gigabit Ethernet and a high-bandwidth backplane leveraging Peripheral Component Interconnect Express (PCIe).

A baseboard, designed to ISO 26262 ASIL D standards, serves as the gateway to the vehicle. “Think of the baseboard as the safety- and automotive-interface for all the drive-by-wire functionality in the vehicle,” Anderson explained. “By adding expansion boards to this base-board, we add higher levels of compute functionality.”

On-road development is being conducted in California with a fleet of test vehicles retrofitted with the new compute platform and equipped with ADAS sensor arrays consisting of lidar, radar, and multiple high-resolution cameras.

“The public-road testing that we’re doing now is between SAE Level 2 and SAE Level 3 functionality,” Anderson revealed. “We’re also doing data collection and research for higher levels of driving autonomy.”

DRVLINE’s architecture features plug-and-play flexibility to handle hardware and software upgrades as autonomous driving technology matures. The design intent is “to provide a very solid framework that partners can use to develop applications without being concerned about the underlying technology,” said Anderson.

Although several Samsung engineers are creating control code to address specific autonomous operational requirements, such as perception, the DRVLINE platform development is based on a high degree of supplier-customer collaboration.

“We want to provide solutions, but only where it makes sense to do so,” Anderson said. Because vehicle OEMs want to write proprietary algorithms for their specific product applications, DRVLINE’s open frame-work allows them to easily augment their own control strategies with what Samsung provides to complete the overall software stack.

Anderson’s development team expanded from five to 72 engineers in 18 months. He expects another wave of engineers to be headed to Samsung’s San Jose, California facility this year. “We’re doing what we think is appropriate to aggressively build out this software solution,” Anderson stated.

Samsung recently formed a strategic partnership with Vienna, Austria-based TTTech, a leader in networking and safety controls for ADAS.



Magazine cover
Autonomous Vehicle Engineering Magazine

This article first appeared in the March, 2018 issue of Autonomous Vehicle Engineering Magazine (Vol. 5 No. 3).

Read more articles from this issue here.

Read more articles from the archives here.