Embedded world 2024: SDVs and AI in automotive

At embedded world 2024, automotive suppliers announced several significant advances to support the development of software-defined vehicles (SDVs) and AI to enhance safety, performance and efficiency. The SDV and AI automotive market is expected to reach $700 billion by 2034, which is about 20% of the global car market, according to a new IDTechEx report. This translates into a 34% compound annual growth rate over the forecast period of 2023 to 2034.

The market research firm attributes the significant growth to improvements in hardware capabilities, in-car connectivity and a shift in customer preferences toward subscription models, such as monthly connectivity subscriptions, in-vehicle payments and one-time software upgrades. However, today, most SDV revenue is derived from connectivity as a service, providing services such as in-vehicle navigation and over-the-air (OTA) software upgrades.

Other key automotive trends cited in the report include the use of AI to enhance features and autonomy levels as well as to develop new vehicles through generative AI and cloud computing. Another trend is using V2X communication to improve safety, efficiency and sustainability.

This year’s embedded world conference showcases some of these new use cases. Here is a sampling of new automotive solutions announced at the show.

Advancing SDVs

The next steps in the evolution toward SDVs are happening with new processors and platforms and a broadening of ecosystems and partnerships. A couple of the biggest challenges are integrating hardware and software and consolidating electronic control units (ECUs) as the automotive electrical/electronic (E/E) architecture continues to evolve.

NXP Semiconductors, for example, is addressing these challenges and more with their new vehicle software platform and latest processors. Launched at embedded world, NXP’s automotive-grade S32N55 processor, the first device in its new S32N family of “vehicle super-integration” processors, is described as a processor that combines scalable real-time and applications processing with hardware isolation and virtualization technologies to support central compute applications. The S32N55 targets real-time vehicle control in SDVs.

The hardware-enforced isolation of vehicle functions enables automakers to consolidate ECUs, streamline software development and support lifetime enhancements and upgrades, NXP explained. The S32N55’s “core-to-pin” hardware isolation and virtualization technologies enables its resources to be dynamically partitioned to adapt over time and be optimized to meet the changing requirements of vehicle functions.

The S32N55 is at the center of NXP’s new S32 CoreRide central compute solution. Thanks to software-defined, hardware-enforced isolation, the S32N55 can host, or consolidate, dozens of vehicle functions with different levels of criticality without interference. This is in comparison to traditional vehicle architectures where core functions like propulsion, dynamics and chassis control have been implemented as discrete ECUs with their own microcontroller (MCU) and wiring, NXP said.

Designed for next-generation SDV development, NXP touts the S32 CoreRide platform as an industry-first vehicle software platform that significantly simplifies complex vehicle architecture development and cuts costs for automakers and tier-one suppliers. It leverages the company’s S32 compute, networking and power management as well as “ready-to-deploy” software from its software partner ecosystem.

NXP said the platform addresses the challenge of moving functions from the traditional multi-ECU to zoned or centralized processing. By leveraging the scalable S32 compute, OEMs can consolidate ECUs and develop flexible architectures, from domain to zonal to centralized, that scale across vehicle classes and generations. The upshot is faster development, while enabling OEMs to focus on differentiation. The first production vehicles using the S32 CoreRide capabilities are expected to ramp up in 2027.

NXP said automotive designers can significantly reduce ECU hardware costs with the S32N55, while the reduction in weight and less material can help extend the driving range and contribute to sustainability. “Ultimately, fewer ECUs and significantly less wiring result in lower manufacturing complexity and time for carmakers,” NXP said.

Key features of the S32N55 processor include 16 split-lock Arm Cortex-R52 processor cores running at 1.2 GHz for real-time compute, which can operate in split or lockstep mode to support different functional safety levels up to ISO 26262 ASIL D; two auxiliary pairs of lockstep Cortex-M7 cores that support system and communication management, tightly-coupled integrated memory and 48 MB of system SRAM. Memory can be expanded with LPDDR4X/5/5X DRAM, LPDDR4X flash and NAND/NOR flash interfaces.

Security features include a firewalled hardware security engine that provides a root of trust for secure boot, security services and key management. Memory error correction and in-line cryptography support functional safety and security requirements.

Other features include an integrated time-sensitive networking (TSN) 2.5 Gbits/s Ethernet switch, a CAN hub for internal routing of 24 CAN FD buses, four CAN XL interfaces and a PCI Express Gen 4 interface.

The S32N55 is sampling to lead customers. NXP and its partners provide a host of board, enablement software, tools and systems support.

Another integrated SDV platform was developed by Infineon Technologies AG and Green Hills Software LLC. They launched an integrated MCU-based processing platform for safety-critical real-time automotive systems. The platform leverages the safety-certified real-time operating system (RTOS) µ-velOSity from Green Hills and Infineon’s new generation of scalable safety controllers, the AURIX TC4x family.

New MCUs are needed to develop ECUs for the vehicle E/E architecture of SDVs, while also providing higher performance and advanced features to meet the requirements of safety-critical systems such as zone control, chassis, radar, electric drives and affordable AI systems, Infineon said.

The combination of the AURIX TC4x family and µ-velOSity RTOS delivers a secure processing platform for OEMs and tier-one suppliers to develop domain and zonal controllers as well as drivetrains for electric vehicles in their next-generation SDV architectures. The new AURIX TC4x devices complement Infineon’s TriCore multicore architecture with a safety and security accelerator suite and Green Hills has ported its µ-velOSity RTOS to the AURIX TC4x family to meet safety requirements.

The Green Hills advanced MULTI integrated development environment (IDE) also supports the processor family. This enables developers to shorten development times and generate the fastest and smallest code for the AURIX TC4x.

The AURIX TC4x uses the next-generation TriCore 1.8 and a scalable accelerator suite, including a new parallel processing unit (PPU) as well as several intelligent accelerators, in areas such as data routing, digital signal processing, radar processing and cryptographic computing. The devices support high-speed interfaces, including Gigabit Ethernet, PCIe, CAN-XL or 10BASE T1S Ethernet.

The Green Hills µ-velOSity RTOS targets the highest functional safety levels (ISO 26262 ASIL D) and is tightly integrated with the MULTI safety-certified tools and compilers.

Green Hills Software also announced a partnership with STMicroelectronics and Cetitec, a Porsche company and software specialist for the development of connectivity and networking system solutions. The companies are collaborating on an integrated and configurable communications platform for use in zonal controllers for the centralized E/E vehicle architecture of the SDV.

The integrated solution manages the communication for consolidation of ECUs in zonal controllers. It’s comprised of Green Hills’ µ-velOSity RTOS, ST’s Stellar Integration MCU platform and Cetitec’s advanced networking stacks, gateways and routing frameworks.

AI in automotive

AI in automotive can be a game changer, making vehicles smarter and safer. For example, AI can be used to improve driver safety for features such as blind-spot detection or even driver attention as well as environmental conditions and vehicle performance. AI is expected to be a key step in the evolution of autonomous vehicles.

Infineon recently announced that the automotive AURIX TC4x MCUs for embedded AI application development passed the AI-specific safety requirements to achieve SAFE AI compliance, as proposed by the Fraunhofer Institute for Cognitive Systems IKS.

The MCUs, with their ASIL-D compliant AI accelerator (PPU), provides a platform for developing embedded AI-based use cases and automotive applications such as motor control, battery management systems, vehicle motion control and siren detection.

The SAFE AI framework based on ISO PAS 8800 and current state-of-the-art AI regulations is an evaluation methodology developed by Fraunhofer IKS to assess the trustworthiness of AI in terms of robustness, data utility, operational design domain (ODD) and environmental conditions. The functional safety measures of the AURIX TC4x family provide mechanisms for compliance with AI regulations and standards at the application level.

“By using the AURIX TC4x family, car manufacturers can assess the safety and reliability of AI solutions and identify potential vulnerabilities during system development and operation,” Infineon said, in a statement. “For safety-critical real-time applications, the use of AI models like neural networks increases accuracy and provides additional safety in conjunction with the existing physical sensor.”

AI in automotive safety applications requires the analysis of massive amounts of data from dozens of sensors throughout the vehicle, increasing the demand for more memory and storage to handle these AI workloads. Micron Technology, Inc., announced that a full suite of its automotive-grade memory and storage solutions have been qualified for Qualcomm Technologies Inc.’s Snapdragon Digital Chassis cloud-connected platforms to power intelligent automotive services.

“Today’s software-defined vehicles and immersive cockpits require highly reliable, ultra-low-latency memory and storage to process massive amounts of data at the edge and enable time-critical near-instant decision-making,” said Chris Jacobs, vice president of Micron’s embedded market segments, in a statement.

Micron’s low-power double data rate 5X (LPDDR5X) memory, Universal Flash Storage (UFS) 3.1, Xccela flash memory and quad serial peripheral interface NOR (SPI-NOR) flash have been pre-integrated into Snapdragon automotive solutions and modules, including the Snapdragon Cockpit Platform, Snapdragon Ride Platform and Snapdragon Ride Flex system-on-chip (SoC). These platforms handle the increasing requirements of workloads for AI technologies.

The automotive LPDDR5X helps improve energy efficiency and performance for intelligent vehicles that increasingly require higher bandwidth and power consumption for applications such as AI-based autonomous driving features. Qualcomm’s latest generation Snapdragon automotive SoCs are the first SoC family enabling the LPDDR5 interface.

The automotive UFS 3.1 storage delivers a boost in speed with two times faster read performance and a 50% improvement in sustained write performance for faster startups and OTA updates. This is particularly important when accessing critical data such as directions from their digital cockpit, Micron said.

The Xccela flash memory is one of the industry’s highest-performance NOR flash memories, Micron said, with a five times the performance and a three times reduction in energy consumption. The quad SPI-NOR flash delivers fast code execution and high reliability for applications such as boot code and program code. “This is especially critical for applications such as booting up a vehicle’s digital cockpit and operating system as soon as a driver starts the engine, as code failure may result in cars being ‘bricked’ or rendered inoperable,” Micron said.

Micron also unveiled the industry’s first quad-port automotive-grade SSD that interfaces with up to four SoCs for centralized architectures in SDVs. Addressing the industry’s demand for a centralized architecture, the 4150AT SSD is built with 176-layer TLC NAND for enterprise-grade performance and application-specific features. It also provides data-center-grade security features and self-test capabilities for in-vehicle detection issues. The company is now sampling the automotive-grade 4150AT SSD.

Automotive sensing

Sensors are key to automotive safety systems and one of those applications is occupant detection. This safety feature has gained significant interest over the past several years with news reports of children being left in the car and dying from heatstroke.

Novelda AS demonstrated a new radar-based occupant detection solution at embedded world. The company has enhanced its ultra-wideband (UWB) in-cabin sensor with a new multi-target seat occupancy detection function. The X7 radar chip is already capable of presence detection, child presence detection (CPD) and vital signs monitoring, adding seat occupancy detection capability through a software upgrade.

UWB can sense through car seats and other materials through centimeter wavelengths, enabling simpler mechanical integration and lower overall system cost, Novelda said. A single UWB sensor can detect human presence in each seat within the car cabin.

The seat occupancy solution can differentiate between people and objects, unlike traditional weight sensors, preventing false seat belt alarms due to objects placed in seats. This is achieved by sensing a person’s tiny motions, including breathing and heartbeat, even if a person is motionless or there is a baby in the seat, Novelda explained.

The X7 SoC has the potential to meet multiple use cases, including ultra-low-power presence and intrusion detection, child presence detection, seat occupancy detection, vital signs monitoring and gesture recognition. In addition, with its heartbeat detection functionality, the chip could detect car crash survivors as well as prevent pediatric heatstroke.

The X7 features a field of view of nearly 180 degrees, which allows companies to reduce the number of sensors in the vehicle for a significant cost savings compared to other solutions, including 60-GHz radar, Novelda said.

Also, the unique UWB sensor technology consumes less than 50 microwatts in a one frame-per-second configuration, outperforming other existing radar ICs on the market today, including UWB or 60-GHz SoCs, the company added.

While security and functional safety are big challenges in the automotive industry, there also is a drive to replace mechanical buttons with touch-enabled surfaces for a cleaner cockpit. Infineon has developed an automotive MCU family that addresses this requirement.

The new Infineon PSoC 4 HVMS family integrates high-voltage features (12 V-regulator and LIN/CXPI-transceiver) with advanced analog features (CAPSENSE and inductive sensing) for touch-enabled human machine interfaces (HMIs) with touch buttons, sliders and touchpads for controlling HVAC, interior lighting, power windows/sunroofs or in door handles. These MCUs are also ISO 26262 compliant, ISO 21434 ready and AEC-Q100 qualified.

The PSoC 4 HVMS also can be used in steering wheels for touch sensing as well as safety-critical hands-off detection. The CAPSENSE module also supports proximity detection for occupant detection or foot kick control. They also are used in generic sensing applications (such as liquid level sensing, Wheatstone bridge sensing, etc.) or in simple actuators such as a PTC heater or interior/exterior lighting, Infineon said.

The PSoC 4 HVMS family, in QFN packages with wettable flanks, is based on ARM Cortex-M0+ processors with up to 128 KB of embedded flash and 16 KB of SRAM. The MCUs can be powered directly from a 12-V battery and include LIN and CXPI PHY. For capacitive sensing applications, the device supports the latest 5th generation CAPSENSE technology with eight times better signal-to-noise ratio (SNR) than the previous generation as well as support for high parasitic capacitance up to 3,000 pF and overlays up to 18 mm. Other features include a 12-bit SAR ADC, up to two operational amplifiers and low power comparators.

Software support includes Automotive Peripheral Driver Library (AutoPDL), Automotive Middleware Library for CAPSENSE, and the Safety Library (SafeTlib) for Automotive PDL. The PSoC 4 HVMS software package is ISO 26262 compliant and developed as a Safety Element out of Context (SEooC) for applications with safety targets up to ASIL-B. The ModusToolbox software development platform will be available soon.

Samples of the PSoC4 HVMS controllers are available for both the 64 K and 128 K families. The series will enter production in 2024.

V2X communications

Vehicle-to-everything (V2X) communications has gained tremendous momentum to improve automotive safety and efficiency in real-time communications between vehicles, infrastructure, networks and pedestrians, along with improvements in technologies for connected cars.

Autotalks, a provider of V2X communication solutions, and Secure-IC, a global provider of end-to-end cybersecurity solutions for embedded systems, have partnered to develop a secure V2X communication chipset that increases the cybersecurity of V2X technology to withstand cyberattacks, data breaches and unauthorized access.

The companies demoed a working module equipped using Autotalks’ TEKTON3 chipset and Secure-IC’s Securyzr hardware security module (HSM) at embedded world.

To ensure that its third-generation chipsets offered robust security, Autotalks licensed Secure-IC’s Securyzr IP to integrate into its chipsets. Together with Autotalks’ V2X Security firmware it creates a certifiable V2X HSM.

In addition to providing cryptographic, randomness and secure processing capabilities for Autotalks’ firmware, Secure-IC’s hardware IP includes certification services, including the Common Criteria framework according to the EU V2X Protection Profile.