What is the development of automotive electronics? Frontier automotive electronics technology at a glance

What is the development of automotive electronics? How do you play it? From advanced sensors to artificial intelligence, cars are fast becoming the paradise for the latest electronics technology and products...

A range of new technologies have emerged in automotive applications, including improvements in power systems, very complex telematics, and autonomous driving. Today's cars have more electronics. However, as features such as Advanced Driver Assistance Systems (ADAS) become standard, rather than expensive options, more advanced functional modules will enter the homes of ordinary people.

Some changes are being quietly implemented by improving sensors, processors and memory, software, and even human-machine interfaces that require real-time integration (see Figure 1). Here are some of the latest technologies and how their relationship with other technologies makes them even more important in automotive applications.

Figure 1. Multiple overlapping sensors are needed to provide the system with context-aware information to implement secure ADAS support

Automotive applications benefit from cameras that can stream 4K video. Together with machine learning (ML) software in artificial intelligence (AI), high-definition cameras are being used for obstacle and object recognition in advanced ADAS applications. Here, higher resolution is critical and it is also useful for backup cameras.

Multiple camera combinations can be used to provide a bird's eye view of Renesas Electronics around the car. For example, the R-Car Development Kit combines video streams from four cameras into a 360-degree view (see Figure 2). This is very useful when parking or navigating in a narrow place. More advanced ADAS systems highlight areas of conflict that may be encountered.

Figure 2. Renesas' R-Car SoC is capable of producing a 360-degree aerial view around the vehicle by weaving video streams from four cameras together.

Recently, there have been significant improvements in two range sensors, LiDAR and phased array radar. These are not new technologies, but significant advances in miniaturization and cost reduction will affect the timing and location of these systems.

For example, Innoviz (see Figure 3), LeddarTech, Quanergy, and Velodyne are companies that offer 3D solid-state LiDAR systems. These systems are suitable for use in fields such as robots, which are getting smaller and smaller and will have multiple units hidden around the car.

Figure 3. Innoviz is just one of many vendors offering 3D LiDAR technology.

InnovizOne has a depth of 200 meters and a range of better than 2 cm. It maintains a field of view of 100 degrees to 25 degrees and a spatial resolution of 0.1 to 0.1 degrees. The device delivers 25 frames per second at 3D resolution of over 6 Mpixels/s.

Phased array radar overcomes many of the limitations of LiDAR, allowing it to operate in rain and snow, otherwise it may mislead the optical system. Radar can be used to compensate for LiDAR and image systems. Some companies are working hard to provide technology in this area. For example, Texas Instruments' single-chip millimeter-wave sensor mmWave handles 76-81 GHz sensor arrays for sensors and ADAS applications.

All of these technologies are used in many areas, from manufacturing to security, and even 3D scanning and printing.

Currently, AI and ML are emerging because they provide efficient image recognition for ADAS, which is critical for safe self-driving or enhanced driving experience. The underlying technologies of AI and ML are based on deep neural networks (DNN) and convolutional neural networks (CNN).

Even in automotive applications, neural networks do not replace conventional software applications, but only solve some of the more difficult problems. Combined with new hardware, they can also be implemented in real time, which is necessary in applications that require high security, such as autonomous driving. In this case, the multi-core processor will play an important role, but the GPU can achieve better results (see Figure 4). Custom hardware is best, such as a dedicated digital signal processor (DSP) that can handle machine learning tasks.

Figure 4. Drive PX2 from Nvidia is the latest multi-core CPU/GPU solution for automotive applications

The parallel processing performance of these solutions performs well in terms of multicore and transistor count growth in the design, even if the upper clock frequency peaks. Customized solutions have lower power consumption than more traditional processor solutions.

Advances in in-vehicle infotainment (IVI) systems are changing the way drivers and passengers visualize and how to connect smart devices and cloud-based applications to their cars. All car manufacturers offer cellular-based Wi-Fi car hotspots. More choices require a more powerful and open approach. In this regard, the GENIVI Alliance promotes open standards that are not related to the operating system.

For example, the Linux Foundation's Automotive Grade Linux (AGL) is an IVI system that has been supported by a wide range of vendors. Toyota's 2018 Camry (see Figure 5) and the future Toyota will use AGL.

Figure 5. Toyota's 2018 Camry will run automotive-grade Linux (AGL) for its in-vehicle infotainment (IVI) system

Given the amount of information generated by a large number of sensors, as well as the data processed and generated by the AI ​​system and the video streams flowing in the in-vehicle network, the number of applications and tasks running on the car system can be surprising. In critical security areas, managing data distribution can benefit from standards such as the Object Management Group (OMG) Data Distribution Service (DDS), which provides secure, real-time management data exchange capabilities across the system. This approach is better than a point-to-point solution in a design that requires less connectivity between applications.

The hypervisor is another common tool, but it is not in the vehicle control settings. However, this change is also changing significantly due to the number and complexity of multi-core solutions and the need to combine safety and safety critical components with IVI and non-critical systems. Automotive-oriented hypervisors are available from vendors such as Blackberry QNX Hypervisor, Wind River VxWorks, Green Hills Software INTEGRITY MulTIvisor (see Figure 6), and Mentor Graphics' embedded management programs.

Figure 6. Green Hills Software's MulTIvisor provides virtual machine isolation for the Type 1 hypervisor, which is becoming more and more in a car environment that carries security and critical subsystems on the same hardware as the IVI subsystem. universal

The hypervisor allows partitioning of virtual machines (VMs) to partition security and security authentication. This means that non-critical components do not require the same level of authentication, which takes time and is very expensive. Similarly, the addition of limited security and security-related components by third parties has become more common in IVI over time.

Like Blackberry's QNX Hypervisor 2.0, such a Type 1 hypervisor is designed to be small, with low memory and performance overhead, but the functionality is critical to performance and security (see Figure 7). QNX provides a priority-based virtual CPU (vCPU) with a configurable scheduling policy. The hypervisor is based on the QNX SDP 7.0 RTOS, providing fine-grained management and security. The QNX Neutrino RTOS is a VM alternative that requires secure authentication. QNX OS for Safety has been certified to ISO 26262 by ASILD and IEC 61508 SIL3 and has been used in the SIL 4 certified EN 50128 system.

Figure 7. Security penetrates into every aspect of the automotive environment. Coordinating and supporting this infrastructure from manufacturing to secure airborne updates can be a challenge, and BlackBerry is trying to provide a complete solution.