(Original title: One of Intel's many acquisitions in the driverless area in recent years on driverless solutions.) Reporter Wilde reports from Beijing In the unmanned area, Intel’s acquisition of Mobileye for US$15.3 billion has become the focus of the industry. In fact, this is just one of Intel’s many acquisitions in the driverless field in recent years. It is transforming from a chip computing giant to the only data company in the world with "end-to-end" capabilities, and smart driving is one of the important areas of its layout. Xu Weijie, Intel’s director of marketing for Intel’s driverless business unit, emphasized in the 21st Century Business Herald that in all companies currently involved in driverlessness, Intel is the only company that can provide “end-to-end†driverlessness from cars, networks, and clouds. Solution company. "China is the world's largest automotive sales market, which means that driven by technology, China's driverless market will grow very fast. This is a market that we attach great importance to." Xu Weijie said. Newly established dedicated driverless division "21st Century": Intel's business is moving from the PC side to the car side, do you think the main difference between these two types of markets? At what stage is your transformation now? Xu Weijie: The transformation has not started in the last year or two. We have already established a special automotive electronics department as early as 780 years ago. In the transformation process, we will still retain and strengthen the traditional business from the PC to the server, but in the automotive electronics sector, we asked the most professional people in the automotive industry to specialize in automotive products. In the automotive electronics field, starting with the infotainment system, we have cooperated with BMW and Lexus; later the products have been extended to integrate products such as cockpits and meters. At the same time, we have newly established a driverless car division and acquired many professional companies. Today, we have very strong technical advantages and products in all aspects of automotive electronics. We work closely with tier1 (first-tier manufacturing suppliers, the same below) companies of global automakers. Some of them have used our products, and others have jointly developed products with us. "21st Century": What is the relationship between the automotive electronics division and the driverless division? Xu Weijie: The driverless business unit was born out of the automotive electronics division and began to be independent from last year. The automotive electronics business is not currently a separate business unit and is placed under the Internet of Things division. These two types of services are closely integrated, but there are differences in the R&D end of the product. For example, automotive electronic products such as cockpits do not require much computing power and involve many contents of the Internet of Things. Driverless products You need very strong computing power, server center, and 5G technology. Full coverage of product lines "21st Century": What are the differences in your strategic positioning between companies such as Nvidia and Qualcomm? Xu Weijie: Overall, from the current market, some companies provide driver-sensing parts and deep learning, some provide 5G communication modules, and more are still in the entertainment cockpit or part of the car network. Intel's characteristic is that these products are available. In the complex system of unmanned driving, end-to-end solutions from the automotive, network, and cloud can be provided to users, helping customers reduce the difficulty or risk of R&D. Functional safety is very important in unmanned vehicles. The ISO26262 standard is very high. There is hardly any company that makes complex chips today to provide CPUs to achieve this standard decision. However, Intel is better able to meet functional safety on our products. We bought Yogitech. All our products on the driverless car solution roadmap support functional safety. In addition to building its own unique capabilities, Intel also expands its capabilities by including cooperation with third parties. "21st Century": Although your product line is very comprehensive, but Qualcomm and Nvidia will also catch up through the acquisition, where are your core barriers? Xu Weijie: In various technologies of automotive electronics, for example, we talked about Mobileye at the level of perception. Our integration after acquisition is still in progress. It is not convenient to talk too much, but you can see that Mobileye has a very strong advantage There are already thousands of different brands of vehicles equipped with its products that have been ran on the road to collect these data, and these advantages are not catching up in one day. "21st Century": In addition to Nvidia and Qualcomm, which end-to-end solutions do you think are more competitive? Weijie Xu: In addition to Intel, we have yet to see other more competitive companies. "21st Century": How much do you plan to invest in chips for driverless cars? how is the progress? When can you expect to mass production? Xu Weijie: I don't know the specific R&D investment data. The production schedule for these chips is inconvenient because it involves different product roadmaps. However, most automotive customers will mass-produce L3 (conditional automated driving) products by 2020, and we will adapt to L3 chips. It is necessary to supply before this. Many manufacturers are likely to release L4 (Advanced Autopilot) and L5 (Fully Automated Drive) grades from 2023 to 2025. Some of the more aggressive manufacturers say that I may reach L4 and L5 by 2020. We will also follow this industry trend. Think that driverless technology is not a gradual development "21st Century": Do you think the technological evolution of unmanned lines is gradual or sudden? Xu Weijie: This depends on different manufacturers' choices. For example, Baidu and Google (microblogging) directly make L4 and L5 products. Many traditional car manufacturers are doing L3 products. We believe that it is not a gradual process from a technical point of view. "21st Century": At the perceptual level, you acquired Mobileye, and the camera-based technology route is relatively a small player in the entire sensing process. The more mainstream market is still dominated by radar. How do you see these two technical routes? Xu Weijie: I personally think that different manufacturers may have their own ideas, but the ultimate trend will be the integration of different technologies. If we say L2 (partially automated driving), L3 level driverless, it may be that Mobileye or other manufacturers' technology is sufficient today, even if you do not need radar through the camera. However, as far as the L4 and L5 levels are concerned, it is possible that the radar cost will be very low with the development of technology, and the effectiveness of the radar and the coverage of the scanning may have a greater advantage. It does not rule out that more manufacturers will also use radar. , But even so, the camera I think is indispensable. "21st Century": Is it now the cost of radar or the cost of cameras? Xu Weijie: Now radar has a very low mass production rate and high costs. However, if the cost of radar can be reduced to less than US$500 or even lower, then the use of radar will not be ruled out. It depends on the technical direction and integration scheme of the depot. "21st Century": You have acquired Mobileye, which is specialized in cameras. Will there be any layout in terms of radar? Xu Weijie: We provide end-to-end solutions. The most important ones are analysis, decision-making, and communication. The outer part is the perception of outside data. The part that can come in to us is mainly to see what Tier1 should use. In this way, we will communicate with various Tier1 manufacturers, including radar manufacturers, and build an ecosystem around our target program. There may be various levels of manufacturers in this ecosystem. In the end, if there is an automobile factory that needs a complete set of solutions, it may be in our ecology. In addition to providing end-to-end solutions, the radar is the manufacturer's own, and the camera is also the manufacturer's own. You can also join together to do it. Intel does not have to buy a radar company by itself. This is not to say that because we bought Mobileye, we do not care about radar. Our point of acquiring Mobileye is not its camera technology, but its image algorithms and data, because these things can also be used in future radars. "21st Century": How do you and LG jointly research and develop 5G communication technologies that can test cars? When can you expect to mass production? Xu Weijie: I cannot tell you the details of the cooperation with LG. However, Intel has progressed smoothly in driverless 5G. The world's first automotive-grade 5G Modem chip in the unmanned field has already been prototyped at the Mobile World Congress. In the future, the various application forms and related standards in unmanned 5G are still not clear. It is still in an evolutionary process. We expect to demonstrate more productization of 5G at the Korean Winter Olympics because South Korea has promoted this earlier.