How have automotive sensors, cables, storage, connectors, and testing solutions evolved?

When it comes to trends such as autonomous driving, ADAS, or car networking, electrification, and intelligence, what we discuss most on a daily basis is probably the innovation of the technology itself and the impact of these technologies on the entire vertical industry of the automobile. When we talk about the fact that the internal software code of high-end cars will reach 200-300 million lines in the future, we will "speculate" how these trends will have a revolutionary impact on the entire structure of the automobile industry.

For example, the industry has changed from a pyramid-shaped hierarchical structure from vehicle manufacturers to suppliers at all levels to a flat horizontal structure, and more participants who were not originally in the automotive industry have joined the industry. In our opinion, these changes are actually complex and "revolutionary" enough. However, at the Aspencore Automotive Electronics Forum this week, we heard another interesting interpretation: Mr. Ye Hai, Senior Strategic Planning Manager of Huayu Automotive Systems, had this view on the development trend of the intelligent connected vehicle market:

Let's first define a word: robocab. Judging from the word structure, it seems to mean a robot taxi. Its actual meaning is a self-driving taxi without a driver. If we ask this question: If every trip in the future can be made by robocab, and the cost of a single trip is lower than driving yourself, would you still be willing to buy a car?

The answers to this question vary from country to country. The proportion of people who answered "yes" and "no" in different countries is shown in the figure above. In any case, this answer at least shows one problem: if the technology is advanced enough, half or even more than half of the people are unwilling to drive themselves. This robocab model is now more well-known as online ride-hailing, but now online ride-hailing still has drivers driving it, while robocab may be an online ride-hailing without a driver.

It generally refers to shared travel, and perhaps this model also exists in other forms, such as time-sharing rental. But they all point to the fact that perhaps in the future we will not only not need to drive ourselves, but also not need to buy a car. We define this service as MaaS (Mobility-as-a-Service).

Judging from the data provided by Ye Hai, China is particularly prominent in reaping the benefits of this trend: "The overall scale of the shared travel market will continue to grow rapidly, with a compound annual growth rate of 19.61% from 2018 to 2030." The four major driving forces of this trend include China's policies (policy support for the development of the shared travel market), capital (various subsidies), demand (increased demand for shared travel services among residents, including some objective factors such as accelerated urbanization, license plate restrictions and tight parking spaces), and technology (technical advances in core components and power batteries, and reduced costs of shared cars; the Internet, 5G, and autonomous driving technologies drive the development of the shared travel industry).

If this trend or prediction is true, then the structural adjustment of the entire automotive industry will not be as simple as we discussed before. On the one hand, industries that were originally unrelated to the automotive industry will also be incorporated into the automotive industry chain system - this is something we have mentioned more than once in previous articles; on the other hand, the industry structure originally dominated by OEMs will be dominated by third-party platforms, such as online car-hailing platforms like Didi, or SAIC EVCARD time-sharing rental platforms.

The second point is probably the most impactful one. In Ye Hai's words, "OEMs will be reduced in dimension in the future; value will be transferred. Companies with fleet management capabilities will have a great say in the industry in the future. For example, if Didi wants to build 2 million cars, this demand will be met immediately." At this time, the car itself becomes an Internet terminal, and it will no longer be essentially different from other Internet terminals.

Then the business model of OEMs will have to be restructured. "The business model of OEMs will change from selling cars to providing value-added services and shared services, and the proportion of revenue from buying car products will decrease." The distribution of the industry's profit pool will change greatly. As shown in the figure above, MaaS will get 34% of the market profit after 2030. In addition, Tier 1 will face the big problem of shrinking market size. At the same time, as we said before, the industry will shift from a pyramid structure to a flatter horizontal structure, so Tier 1 will begin to be impacted by cross-border companies that are not originally in the automotive industry, such as software companies.

When the penetration rate of the shared travel market reaches a certain level and the utilization rate of social vehicles increases, the total demand for cars will begin to decline. "It is expected that the social demand for vehicles will decrease by 30-50% from 2035 to 2040." Then the changes at all levels mentioned above will be further strengthened. It is not difficult to understand the structural changes in the above figure, and how the "Vehicle buyer" at the top level will change the industry.

This is an interesting deconstruction, and it seems very bold. Regarding the trend of MaaS, Ye Hai's suggestions for Tier 1 manufacturers include: 1. Pay close attention to the market opportunities of MaaS, discuss PMV (Purpose-built Mobility Vehicle) cooperation opportunities with travel service providers and OEMs, seek to expand the share of single-vehicle matching, and maintain business growth - this seems to be the current multi-faceted efforts of HUAYU Automotive in all levels of intelligent systems, intelligent form systems, intelligent body and cockpit systems, and intelligent power and energy management systems.

2. With the changes in the modules that make up the car, Tier 1 suppliers need to explore market opportunities based on new platforms. Volkswagen divides the software platform into five parts, namely "operating system and interconnection", "intelligent body and cockpit", "autonomous driving", "vehicle and energy performance", and "service platform and travel services". This actually corresponds to the above-mentioned directions of Huayu Automotive. 3. Increase the penetration rate in the Chinese market. The Chinese automobile market is the largest market in the world. "In 2018, the growth rate of shared cars exceeded 300%, and the growth rate of online car-hailing was 88%." This responds to the rapid development of the MaaS model in China in the future.

Improve ADAS capability layout: sensors, cables

However, before discussing such a conceptual trend, let’s talk about more specific issues. In a relatively specific suggestion, Ye Hai mentioned “through strategic alliances, joint ventures, acquisitions, etc., to improve the ADAS capability layout, including radar, camera, lidar, chips, data fusion, etc., to form the ability to provide driver assistance/autonomous driving system solutions at all levels.”

This is actually the industry trend that is happening now, especially the integration and fusion of multiple solutions. We mentioned in " Building an Era: 10 Trends in Automotive Electronics and Software Architecture " that in the short and medium term, the number of sensors in the car will soar. But in the long run, the increase in the number of sensors will actually lead to an increase in material costs, and how to reduce costs is still a long-term trend. External sensors, camera + radar fusion solutions, will inevitably dominate the market.

At the Automotive Electronics Forum, Li Jianlin, General Manager of Nava Electronics, mentioned the integration of millimeter-wave radar and 5G C-V2X, and also involved the integration of camera vision solutions. "The combination of vision and millimeter-wave radar will be the mainstream solution for a long time in the future. The advantage of vision (camera) is that it works well in 70%-80% of cases, but once it leaves the light source, vision becomes a shortcoming. Millimeter waves are air waves and are not much affected by air and external conditions. The combination of vision + millimeter-wave radar forms an all-weather solution, which is also the mainstream in the market."

Especially when millimeter waves are used for forward radar, radar does not have enough information to judge many tasks and needs to be integrated with vision. "In 2018, our Ministry of Transport stipulated that buses longer than 9 meters and 18-ton trucks must be equipped with FCW (forward collision warning system) after 24 months. In addition, there is a requirement to support WiFi and Bluetooth, which has also been highly recognized in the industry." This actually promoted the development of millimeter wave radar.

"So many domestic companies have begun to engage in this field. Although the demand in this market is huge, the current mainstream market is still occupied by foreign imports. So in the next period of time, what we have to do is import substitution." Li Jianlin said.

Li Jianlin explained to us in detail the various "pitfalls" encountered by NanoWave Electronics in the manufacturing process of millimeter-wave radar modules. "For example, the black shell above is made of special wave-transmitting materials, including some metal media fused in the shell. We spent more than two years on this shell alone, so the cost is not as low as imagined." And in the signal processing process, there are various experiences summarized by NanoWave Electronics when making MCU selections.

Therefore, in the overall solution, "MIMO antenna technology is combined with the ESPRIT super-resolution algorithm, and complex debugging is used to improve anti-interference performance; digital beamforming and single pulse deambiguation are used to improve angle accuracy and resolution; the DBSCAN clustering algorithm is combined with the nonlinear Kalman filter tracking algorithm to accurately calculate the target track."