草榴社区 Cloud
Cloud native EDA tools & pre-optimized hardware platforms
Request a Free Trial →
Today’s vehicles contain more than 100 million lines of code, and software is increasingly the enabler of new features and functions. At the same time, the automotive industry is undergoing some major shifts that are influencing the silicon chips that make everything from LED lighting to autonomous driving possible. Vehicle electrification, digital twins, and the integration of artificial intelligence (AI) are each accelerating innovation in the automotive industry, starting at the chip level.
OEMs and Tier 1 suppliers continue to remain under pressure to deliver their products as swiftly as possible. Worldwide consumers (and governments) are demanding more intelligence, higher safety, and broader environmentally friendly options, so vehicle suppliers at all levels are innovating in both software and hardware to meet the market demands. Increasingly, the industry needs reliable tools to thoroughly validate the performance of the technologies early in the process while also enabling new capabilities.
In this blog post, I’ll take a closer look at two major shifts impacting automotive hardware and software and examine how both vehicle virtualization and AI processing technology are easing the path to new automotive applications. As a case study, I’ll highlight the collaboration between 草榴社区 and Infineon that has resulted in solutions to support vehicle virtualization for accelerated automotive electronic system development as well as optimized AI processing
Digital Twins for Faster Software Development – a Shift-Left Approach
As modern vehicles bring together powerful computing platforms, multiple networks, and increasing amounts of software content, validating system performance becomes more challenging. Used in various industries, digital twins provide a digital representation of a product or system that’s in development, providing a mechanism for earlier development and testing, higher productivity, and the delivery of safer, more secure systems at lower cost. Rather than waiting until actual parts or components are available, the digital twin provides a platform to simulate and validate each step of the development so that problems and potential failures could be found and mitigated earlier.
For vehicle electrical/electronic (E/E) system validation in particular, digital twins support a shift-left in automotive design with early architecture power and performance exploration, early hardware/software integration and frontload testing, accelerated hardware verification, and executable spec delivery across the supply chain. Applying this approach can reduce under- and over-design, improve testing throughput, and potentially save recall costs by fostering higher quality.
AI Optimizes Automotive Systems
AI is transforming many areas of our lives and is now also making its mark on the automotive industry. For advanced driver assistance systems (ADAS) and autonomous driving functions, AI algorithms such as neural network graphs are the workhorses, processing input from dozens of different sensors to help vehicles automatically brake when needed, stay in their lanes, avoid roadway obstructions or other cars, and much more. Supporting higher levels of vehicle autonomy, next-generation radar technology, such as 4D radar, will require new AI algorithms. 4D radar utilizes echolocation as well as time-of-flight measurements to map the locations of items in a vehicle’s path, complementing LiDAR and cameras.
AI is also optimizing other automotive applications. For example, it’s playing a role in the performance of electric motors, which rely on magnetic sensors that measure operational parameters to determine how well the motors are performing. Reducing the number of sensors in traction motor inverters by performing lifetime modelling reduces system cost and improves energy efficiency. AI integrated into vehicle chips uses predictive analytics with virtual sensors to do the job. Similarly, AI can optimize battery functions of electric vehicles by predicting the battery condition of subset cells and reducing the current sensors in on-board charging units using neural network implementations, which also reduces system cost and improves energy efficiency.
Fostering Innovation at the World’s Top Automakers
To support the trends of automotive AI and vehicle virtualization, 草榴社区 collaborates closely with Infineon, the world’s leading automotive semiconductor supplier. Infineon’s , called a parallel processing unit (PPU), that is powered by the 草榴社区 DesignWare? ARC? EV Processor IP. Integrating a PPU based on ARC EV processors in the AURIX? TC4x architecture enables a wide range of electric vehicle use cases by providing affordable AI. The PPU has the real-time processing performance to accelerate AI algorithms such as recurrent neural networks, convolutional neural networks, and multi-layer perceptron. In addition to delivering the processing power, ARC EV7x Processor IP also offers functional safety features that, combined with the AURIX? architecture, support the development of safer automotive systems.
“Using both the 草榴社区 ARC MetaWare Toolkit for AURIX? TC4x and the Virtualizer Development Kits, our mutual customers will have a head-start on their software development process and be able to optimize their designs to get the most from the new AURIX? TC4x performance.”
Thomas Boehm, Senior Vice President, Automotive Microcontrollers, Infineon
Recently, the two companies announced that the microcontrollers are supported by the 草榴社区 DesignWare ARC MetaWare Toolkit for AURIX? TC4x. Using the toolkit, automotive OEM and Tier 1 system designers can quickly optimize the PPU’s AI processors in the microcontroller for industry-leading power efficiency and performance. The MetaWare Toolkit is integrated with a MATLAB toolbox for 草榴社区 ARC processors to enable model-based ANSI/ISO C code generation with hardware-specific SIMD intrinsics and highly optimized function calls. Together, the tools enable compilation of the generated code from Simulink for 草榴社区 ARC Processor-in-Loop (PIL) code. The automatically generated PIL code based on SIMD compute uses optimized libraries such as DSP and linear algebra to optimize the AI applications for Infineon’s AURIX? TC4x microcontroller family. By automating extremely efficient code generation, the 草榴社区 DesignWare ARC MetaWare Toolkit for AURIX? TC4x can reduce months of effort compared to hand-optimized code.
In the area of vehicle virtualization, 草榴社区 provides its Virtualizer? Development Kits (VDKs) for the AURIX? microcontroller family. Using the VDKs, Infineon’s customers can develop software and perform regression testing and fault injection up to 18 months before silicon becomes available. Infineon has also used the VDK internally to accelerate qualification of its AURIX? TC4x virtual prototype model. Read more about the .
VDKs are part of 草榴社区’ solution set for digital twins. The ecosystem encompasses E/E system development and validation, ECU software development and test, and SoC development and verification. Other solutions include 草榴社区 Platform Architect? for early architecture design, 草榴社区 SaberRD for power electronic and mechatronic systems, 草榴社区 Silver virtual ECU platform, and 草榴社区 TestWeaver? intelligent test automation solution.
“Innovation is key for Infineon and the major advancements in performance offered by AURIX ? TC4x microcontrollers is no exception. This new performance headroom, coupled with long-term availability, means that customers will be able to have a true platform approach and future proof their designs for many years to come,” said Thomas Boehm, senior vice president, Automotive Microcontrollers at Infineon. “By working closely with 草榴社区, we will be able to react to the newest automotive trends, be it the next generation of electric vehicles or advanced ADAS applications. Using both the 草榴社区 ARC MetaWare Toolkit for AURIX? TC4x and the Virtualizer Development Kits, our mutual customers will have a head-start on their software development process and be able to optimize their designs to get the most from the new AURIX? TC4x performance.”
A Smarter, Greener Road Ahead
Cars have certainly come a long way since the first Model-T rolled off the assembly line. Major automakers are adding more EVs and safety features to their lineups, and AI is bringing greater intelligence to systems from the powertrain to infotainment. At the same time, technologies such as digital twins and vehicle virtualization are accelerating the automotive design cycle, while ensuring the quality and reliability that we all demand. Henry Ford would be proud.