Elektrobit, Nvidia, and Infineon team up to tackle safety systems

  • 21-Oct-2015 11:20 EDT

Elektrobit, Nvidia, and Infineon are helping developers enhance their ADAS designs.

Advanced driver assistance systems (ADASs) are becoming increasingly complex as developers add capabilities and enhance sophistication. That’s prompted Elektrobit, Nvidia, and Infineon Technologies to team up and offer an ADAS solution that can serve as a development platform for autonomous driving. The solution consists of the Nvidia Drive PX self-driving computer, Infineon’s AURIX 32-bit TriCore, and Elektrobit’s AUTOSAR 4.x-compliant EB tresos software suite.

Nvidia Drive PX makes it possible to analyze and combine multiple HD camera and sensor inputs, while the Aurix real-time processor has enhanced embedded safety and security features. Elektrobit’s tresos software facilitates the integration of Linux and AUTOSAR applications and enables cross-CPU-communication at high safety integrity levels. The companies are addressing many trends, including functional safety, increased computing power, versatile communications, and redundancy.

“Our software runs on a board with two Nvidia Tegra chips and an Infineon Aurix TriCore chip,” said Karsten Hoffmeister, Director of EB’s autonomous initiatives. “The board has CAN, LIN, Ethernet, and FlexRay communications channels, with redundant Ethernet connection paths so if something goes wrong the other channel will be available.”

Combining multiple processors on a board highlights the drive towards a single electronic control unit (ECU) that analyzes many inputs so it can make driving decisions. Using fewer, more powerful ECUs will also simplify architectures and reduce costs. While engineers attempt to trim the number of controllers, they’re also adding backups to ensure that safety systems don’t break down because a single element fails.

“At the system level, you’re not able to use one single ECU. You need redundancy,” Hoffmeister said. “Right now, the single point of failure is still the power supply. In the future, we’ll see two power supplies.”

The increasing complexity of ADAS is also prompting greater use of AUTOSAR. It lets design teams use software modules for common low-level tasks so they can focus on high-level operations.

“AUTOSAR is really the standard architecture for developing non-infotainment-based parts of the vehicle like brakes, steering, and engine control,” Hoffmeister said. “Instead of having people configure the CAN stack again and again, we’ve done it. With AUTOSAR, the interface will be stable, though all the OEMs have their own flavor, especially in diagnostics and updating.”

Hoffmeister noted that using the standard still requires a fair amount of effort.

“AUTOSAR is complex. We have predefined a lot so it runs easily. They don’t have to spend weeks,” he said. “AUTOSAR is still really for specialists; people using it should really take care to understand it.”

While more OEMs and Tier 1s are using AUTOSAR in production vehicles, Linux is another story. It’s included in Elektrobit’s software suite, but at present, its role will be largely limited to design and testing.

“Linux is very good for development, there are lots of libraries out there,” Hoffmeister said. “You can start out using Linux on a PC; then when you go to mass production you can reduce Linux to certain parts where it works.”

The inclusion of Ethernet also highlights the industry’s growing adoption of commercial technologies. The network permits faster transfer rates and more flexibility.

“Many OEMs are moving to Ethernet for bandwidth,” Hoffmeister said. “Cameras can send Ethernet frames, and it’s good for data. Also, if you offer a service and want to change it later, you have dynamic allocation of services and quality of service.”

HTML for Linking to Page
Page URL
Rate It
4.33 Avg. Rating

Read More Articles On

Euro NCAP will establish a separate category for autonomous vehicles, but there is not likely to be one for cars that are claimed to protect all occupants from serious injury or death.
Unveiled at the 2017 Geneva Motor Show, this uber-performance hybrid uses F1 technology.
Toyota tests ‘relationship-based driving’ along with your friendly, AI-enabled, net-connected clone companion.
Experts serve cautionary tales for connected cars while discussing potential cyber solutions.

Related Items

Technical Paper / Journal Article
Technical Paper / Journal Article
Technical Paper / Journal Article
Training / Education
Training / Education