The PAVE360 simulation platform is on display at the Siemens Center for Practical Autonomy in Novi, Michigan. In an announcement I would definitely classify under “Cool Tech,” Siemens announced a new simulation platform that allows you to design and test a System-on-Chip (SoC) for cars in a completely virtual environment.
Called PAVE360, the new platform simulates an entire vehicle plus the environment and driving scenarios so that you can design and test a virtual chip all the way through its operation, or crash test, before committing to the costly manufacturing of the chip and associated computing platform.
The need for simulation platforms has become critical to avoid both the time and risk of training driver assisted (ADAS) and autonomous vehicles on the roads, but most of the simulation platforms only focus on the training or validation of the entire car once the design is complete. PAVE360 is focused on the design of SoCs, vehicle systems, and even the entire vehicle.
PAVE360 combines the tools from 22 acquisitions at a cost of over US$12 billion with tools developed in-house for a complete hardware and software simulation platform. The platform enables virtual SoCs to be tested with other virtual components in a virtual car driving in a virtual city under virtual weather conditions. In other words, you can simulate everything and change the models throughout the development of the platform to the point of even using some of the hardware components for verification and validation.
The platform also allows the various systems vendors to work together in a secure virtual environment. Components and attributes of the PAVE360 simulation platform from Siemens With a number of SoCs and automotive AI computing platforms available, is there a need for custom SoCs for automotive? If Tesla is any indication, the answer is yes. Tesla recently announced its own custom IA processing chip referred to as the FSD.
While the Tesla SoC will not win any performance crowns, it is optimized for Tesla’s software and system requirements, such as performance, power, size, and heat dissipation. And, according to Tesla, it is saving the company 20% of the cost over the Nvidia platform that it was previously using. TIRIAS Research believes that custom SoCs will become more common as the automotive industry progresses from driver-assisted vehicles to fully autonomous vehicles.
However, not every OEM is going to following the Tesla path of hiring and designing its own SoC. PAVE360 could also be used by current semiconductor providers and semiconductor design houses to build custom SoCs for each auto OEM, auto platform, or car model. A PAVE360 platform is on display at the Siemens Center for Practical Autonomy in Novi, Michigan.
The automotive market is not the first application or segment to develop custom SoCs. We have seen this in smartphones, game consoles, and servers as OEMs seek to optimize and differentiate their products. Automotive is just the latest market segment to consider custom SoCs and the benefits extend well beyond the value of the silicon or compute platform.
With PAVE360, the entire car can be designed and tested in near real time. That means the potential to reduce the design of a vehicle from five years or more down to three years or possibly even one year in the future. Siemens also offers other tools that can aid in the design, manufacturing, and lifecycle management of vehicles and other vehicle components, such as tires and batteries.
While it is overwhelming to think of the increasing processing requirements of future cars as they progress toward fully autonomous vehicles, it is amazing to think that you can simulate everything from the SoC to driving scenarios in a completely closed-loop system. From someone that once designed rockets for General Dynamics Space Systems, I would have done anything to be able to simulate systems at this level. I would imagine that the platform may have other applications as well. Airplanes are the first that come to mind.