SiliconAuto Leverages Siemens PAVE360 for ADAS Software Development
The Stellantis/Foxconn joint venture is writing driver assistance code before the chips it will run on is available.
At a Glance
- Lets coders write software before ADAS chips are available
- Developers can code as the model progresses in fidelity
- PAVE360 can also help debug the hardware once chips are available
Siemens’ PAVE360 software is helping SilconAuto coders write software for advanced driver assistance systems before the chips that will run the software are ready.
SiliconAuto is a joint venture of Stellantis N.V. and Hon Hai Technology Group (“Foxconn”), established in 2023, that is dedicated to developing leading-edge semiconductors for the automotive industry.
SiliconAuto is using PAVE360 to build an automotive standards-based multi-client, multi-fidelity, virtual development environment that mixes existing tooling and models with new virtual SoCs and real-world inputs to gain early insight, enable early software development, and provide key metrics for decision criteria.
The goal is to provide a virtual reference development environment for its new ADAS System on a Chip (SoC) - prior to silicon hardware availability.
“By leveraging PAVE360, we are not just accelerating development but also paving the way for future-ready vehicles that meet the highest standards of safety and performance” said SiliconAuto vice president of sales and marketing, Shawn Tien.
“The pace of development for the advanced technologies required for next-generation electric vehicles is accelerating - OEMs and suppliers need to be able to prove out, simulate and iterate software capabilities long before hardware becomes available,” explained Siemens Digital Industries Software vice president of Hybrid and Virtual Systems, David Fritz.
Such virtual prototypes are nothing new, Fritz said. It is just that their “value is suspect.” That’s because they’ve been slow and inaccurate.
PAVE360, in contrast, lets developers code while progressing in fidelity, starting off with more abstract work that explores the architecture to provide valuable metrics as the work continues.
“Then you can increase fidelity and get more accurate,” he said. “You can get into details of cache sizes and latencies. The software model can run with 100 percent accuracy and get full power and performance out of it.”
The important capability of PAVE360 is that in all fidelities users can run all workloads. “That’s our big differentiator.”
“SiliconAuto had hardware running on our models very quickly,” Fritz noted. “It was a poster child for this whole methodology. We like how aggressive they are. How much more quickly they move than the average western automotive company.”
The visual aspect of PAVE360 illustrates the ADAS system’s decision-making for developers testing the system. “You can see the vehicle make decisions,” he said. “They demonstrated that early and that rolled into perhaps some modifications of how the silicon was going to work.”
And PAVE360 is helpful once the silicon is ready, providing the ability to show why it behaves the way it does if there are questions. “Once the silicon comes back, if things aren’t exactly as one would expect, you say “Let’s run in model and gain insights of what’s happening inside the silicon,” Fritz said. “It will be a very valuable debug tool when they are bringing up the silicon and the software.”
About the Author
You May Also Like