China and U.S.’s Different Approach towards Self-Driving Technology
The reason that the field of autonomous (or self-driving) technology has attracted so much investments and headlines these days is simply that it is not one single industry. The technology itself involves the integration of multiple advanced technologies such as integrated electronic hardware, big data, cloud computing, artificial intelligence, and high-precision maps. The field is actually better defined as a massive ecosystem, one that is going to change not just how to travel from point A to point B, but one that has extremely far-fetched implications on how cities are planned and function.
There are thousands of startups globally that are developing technologies for this ecosystem. But two countries, as we probably expected, stand out from their peers in terms of the having the highest vision of creating the entire supply chain that enables every pieces to eventually fall into place. It is the same old enemies: China and U.S. However in order to fully grasp the potential of the technology, we have to take a step back to understand that the two countries actually take a very drastically approach towards every dime that it is throwing at the tech. Let’s dig into 5 key areas of our analysis: I) dimensions of technology route, II) enterprise relationship, III) leading company, IV) industrial chain, V) R&D environment.
Technical route: connected VS autonomous
The competition between China and the United States in autonomous driving is not only about the higher degree of autonomy of vehicles, but also the right to speak of future industry standards. As far as the technical route is concerned, the two parties can even say that they are singing a drama around this field.
Why do you say that? Because China’s focus is creating network-connected autonomous driving, while the United States is focused on autonomous driving from a single vehicle perspective.
Network-linked autonomous driving relies on high-speed, low-latency wireless communication technology (5G or above) to exchange data between on-board sensors and surrounding information sources (smart traffic lights, smart street lights, etc.) from the infrastructure to realize the network system Control the vehicle.
The autonomous autonomous driving developed by American companies is dedicated to allowing vehicles to approach the level of human drivers through perception and learning, thereby autonomously planning driving routes. The vehicle completely relies on its own sensors and decision-making control systems, and does not rely much on infrastructure such as the communication environment.
Both have their advantages and disadvantages. The advantage of networked autonomous driving lies in its higher theoretical feasibility. As long as the infrastructure is in place, it can be realized, and it greatly reduces the requirements for the intelligence of the vehicle itself. Of course, this is undoubtedly a huge and complex system that requires large-scale supporting construction, and the government investment cost is self-evident.
Autonomous autonomous driving is equivalent to a more thorough and intelligent upgrade of the ADAS advanced driving assistance system on current vehicles. It seems to be closer to us, but it currently faces many problems. For example, the cost of bicycles is too high, and the cost of the after-installation scheme with lidar as the main sensor is as high as hundreds of thousands of dollars; accidents occur frequently, causing many ordinary people in the United States to question or even resist it; relevant laws and regulations are lagging behind, and accidents are punishable.
However, if you look more optimistically, the cost of autonomous self-driving bicycles can be greatly reduced with the expansion of scale and technological progress, and accidents can also be gradually avoided with the improvement of perception accuracy and AI algorithms. Related laws and regulations are such superstructures. Will come because of the development and growth of the economic base.
The L4 self-driving car launched by Google in 2014 Firefly Since the United States started early in this field, the Silicon Valley self-driving companies represented by Google Waymo have a huge advantage in related technology talents, and they are far ahead. The road test mileage, so in terms of the autonomous driving capability of the vehicle itself, the United States is currently well-deserved №1.
It should be noted that although our direction is connected, it is also urgent to improve the intelligent level of vehicles. Therefore, Chinese autonomous driving companies represented by Baidu are trying to catch up with American autonomous driving companies such as Google Waymo.
Even though the U.S. companies had a head start in timing (especially Google Waymo, which was launched more than 10 years ago), China seems to have taken a very different approach after learning from the U.S. experience. The major differences can be summarized below.
1) Long Term Feasibility
First, as mentioned in the previous article, the theoretical feasibility of connected autonomous driving is higher, and one more point is that its strategic significance is also more long-term. Everyone may wish to think about it. China has realized a smart factory with driverless transport vehicles, and a series of concepts such as smart ports and smart mines are not far from the official landing. Is that as long as the scope of the smart scene With step-by-step extension, can the historical feat of smart city be realized logically? And the so-called smart city, isn’t it a future human habitat that uses network systems to efficiently dispatch?
2) Smart City Concept
In the second aspect, the development of connected autonomous driving is a typical strategy for maximizing strengths and avoiding weaknesses.
When it comes to China’s strengths, it is apparent that the connected-network concept has been further developed to a much clearer picture, when compared to the one in U.S. This can be seen in the latest V2X Muti-tier Cloud Infrastructure Whitepaper that the central Chinese government has published together with all the leading companies in the broad field including tech giant Huawei, Alibaba, Tencent, Baidu and others. (This whitepaper has been published only in Chinese language. For those that would like to get an fully English copy, you can send us an email at: email@example.com)
Among the supporting infrastructure, one of the most important technology is 5G. In this regard, China has an absolute lead in the development of 5G communication. V2X is an information interaction platform that connects cars and cars (V2V), cars and roads (V2I), cars and people (V2P), cars and clouds (V2C), etc., including security solutions, electronic and electrical architecture, and cloud platforms. . C-V2X refers to the V2X technology based on cellular communication, which is suitable for LTE 4G and 5G era.
At present, the V2X communication system is mainly based on the DSRC standard led by the United States and the C-V2X standard led by my country. With the popularization of China’s 5G communication, the C-V2X standard surpassing the DSRC standard is just around the corner.
3) Industrial Economy
The third aspect is looking at how the autonomous technology actuallly impacts the broader economy for both countries. According to data from China’s National Bureau of Statistics in 2019, the “second industry” (which includes broadly mining, petroleum and oil shale manufacturing, vehicles, bicycles, aviation, ships, defense, food, meat processing, alcohol, tobacco, textiles, fur, papermaking chemistry, plastics, pharmaceuticals, electronics, semiconductor energy, power, renewable energy, buildings) account for 39.0% of China’s GDP. China’s development of autonomous driving is not only to change the way people travel, but also to provide strong support for the digital transformation and upgrading of the entire country’s industry, as well as reducing costs and increasing efficiency.
In comparison, the United States situation is quite different. Its service industries account for more than 80% of the total GDP. The core motivation for Silicon Valley companies to vigorously develop autonomous driving has a much narrower emphasis on the fundamental economical benefits from the way that the technologies can be monetized.
As we have initially set up the framework for this analysis in the article, the point is not to draw an easy conclusion on who might win the race. Most importantly, it is important for the public to fully understand not only the progress of technology, but also the complex conditions that are in place. These macro-economic conditions will play a very critical roles in determining the race in the next decade. After all, in order to realize safe and reliable autonomous driving at scale, there are so many parties involved including the technology startups, car manufacturers, government and the public.
As mentioned earlier, China is taking a route of connected autonomous driving that relies on a large number of supporting infrastructure. What does this mean? First of all, this is by no means something that one or several companies can accomplish, but requires extremely complex multi-field technology integration; secondly, wherever large infrastructure is involved, government departments will play an important role.
Both companies need to cooperate closely with each other and have government support, so the result is not unexpected, the relationship between companies will form a strong division of labor, and everyone can work together to do something awesome. Ever since, companies that excel in various fields such as autonomous driving chips, autonomous driving AI algorithms, computer vision, Lidar, millimeter wave radar, cameras, high-precision maps, 5G communications, smart traffic lights, smart street lights, etc. are conquering. In this way, a relatively complete supply chain has been formed in China.
Google Waymo, relying on its own technological leadership, and the parent company behind it, not only does algorithms and maps, but also develops its own core hardware such as TPU autopilot chips and high-beam laser radar. In addition, it has been revealed many times that the business model will include the provision of travel services, logistics/self-driving trucks, urban buses, and authorized automakers to use their own self-driving systems.
Google Waymo’s extremely large vision has created the enemies of the entire industry for a while, and it was rebelled by GM, Ford, Uber, Intel and others:
In February 2015, Uber and its parent company Google broke their skins and officially announced that they would enter the L4 autonomous driving field. Since then, they have been digging people from Google, including Anthony Levandowski, the core figure of Google’s autonomous driving, and Otto, the self-driving truck company founded by the latter, was acquired, and he would not hesitate to lose money in a lawsuit.
In March 2016, General Motors was dissatisfied with Google’s “humiliation” and acquired the autonomous driving startup Cruise for $1 billion, which later developed into Google’s Waymo’s biggest rival in the US market;
In early 2017, Ford Motor also acquired Argo AI, founded by another Google “traitor” Bryan Salesky, for $1 billion.
In March 2017, Intel acquired the Israeli ADAS supply chain company Mobileye for $15.3 billion, and later entered into alliances with BMW and Delphi.
All in all, China’s domestic autonomous driving companies express a more ecosystem and collaborative characteristics. But having said that, the reason why American autopilot-related companies can achieve a significant lead in the field of autopilot depends on the early start and strong technical strength on the one hand, and on the other hand, it is also related to this highly competitive market environment.
Leading companies: open Baidu and closed Google
The leading companies in autonomous driving technology in China and the United States are Baidu and Google. There should be no objection to this.
Google Waymo is taking a closed route, committed to making its own autonomous driving similar to the existence of Apple’s IOS system, and then applying it to various types of vehicles and multiple scenarios to build a closed ecosystem entirely dominated by itself , And then as much money as possible into their own pockets.
Baidu Apollo chose the opposite route to Google, using an open ecosystem to make many friends and few enemies, determined to become the “Android” of the future automotive industry.
The core idea of Baidu Apollo is to let more companies and developers use it through open source algorithms and free simulation test platforms, so as to obtain a large amount of data feeding, continuously improve the ability of AI algorithms, and then attract more developers to use and occupy A larger market, so repeated cycles, strive to quickly catch up and even surpass the opponent.
In Baidu Apollo’s business model, it does not earn money for selling cars, travel operations, autopilot system integration, or deep customization of the system. They believe that through cloud computing, 3D high-precision Services such as maps, intelligent voice, and on-board OS systems are sufficient. After all, the Chinese market is large enough, and small profits but quick turnover can make a lot of money.
With such an open and mutually beneficial attitude, Baidu Apollo has initially formed a strong and complementary circle of friends, including traditional auto companies such as FAW, Dongfeng, and BYD, and autonomous driving chip suppliers such as NVIDIA, Intel, and Horizon. Auto parts giants such as Bosch and Continental, lidar suppliers such as Velodyne, Hesai, and Sagitar, travel operators such as Shenzhou Car Rental and Shouqi Car-hailing, and autonomous driving startups such as Zhixingzhe, Momenta, and Neolithic… as of At present, the total forecast has exceeded 200 related enterprises or units.
Cao Xudong, the founder of Momenta, once said that after joining Baidu’s Apollo ecosystem, they can empower the Apollo platform in the field of computer vision. In turn, with the aid of the Apollo ecosystem, they can serve many car companies and do deeply customized autonomous driving systems, so this is a win-win cooperation.
Li Xiaofei, the co-founder of Zhixingzhe, said frankly that using the 3D high-precision map in the Apollo ecosystem can help them create low-speed unmanned vehicles in the park scene faster. Using the Apollo simulation platform will increase their algorithm development efficiency by at least 2 times.
In order to make a clear comparison, we break down the entire value chain into four key parts: 1) perception system, 2) decision-making system, 3) execution system and 4) communication system, according to the composition of the autonomous driving system.
Leader VS Chaser
In general, China belongs to the role of a chaser, and the United States is the leader. China has an advantage in the communication system part (C-V2X based on 5G), while the remaining three major parts are lagging behind, especially the execution system.
Perception systems include cameras, Lidars, milli-meter wave radars, high-precision maps and positioning, etc. The core sensor CMOS and image processor DSP of the car camera are mainly monopolized by foreign companies, including Sony (Japan), Samsung (South Korea), Texas Instruments (USA), Mobileye (USA), etc. In terms of lenses, China’s Sunny Optical has the largest market share in the world, and OFILM is not bad.
In the field of lidar, American companies are far ahead. Velodyne’s 16, 32, 64, and 128-line lidars, and Quanegy’s M8 and S3 all-solid-state lidars are all star products. However, since China has a relatively complete industrial chain and a good foundation of optomechanical technology, companies such as Hesai Technology, Sagitar Juchuang, and Yexing Technology have greater opportunities in this field.
The milli-meter wave radar market is currently basically occupied by foreign manufacturers, such as Bosch (Germany), Continental (Germany), Hella (Germany), Delphi (U.S.) and other companies have deep technical accumulation and practical experience. In China, millimeter wave radar manufacturers such as Suzhou Haomibo, Shenzhen Anzhijie, Nanoradar Technology, and Nanjing Falcon Eye are also actively deploying and trying to catch up with the international level.
As for high-precision maps and positioning, China and the United States can be said to be the in similar positions. The United States has Google Maps, Mapbox, DeepMap, etc., while China also has Baidu Maps, Gaode Maps, and NavInfo.
The decision-making system includes on-board computing platform, autonomous driving algorithm, etc. In-vehicle computing platforms mainly refer to autonomous driving chips. The United States has an ultra-luxury lineup of companies such as NVIDIA, Intel, Mobileye, Texas Instruments, Qualcomm, Tesla, and Google. China’s Huawei, ZTE, Horizon, Cambrian, etc. There is still a big gap.
In terms of AI algorithms, Google Waymo and Tesla are leading the number of mileage tested on the road. On the other hand, China’s Baidu is leading in their mileage of test in China, having recently launched their Robotaxi operations to public in selected districts in Beijing. In addition, there have been a number of fast-growing startups that are progressively developing full stack technologies on the public streets in Chinese cities with the support of the Chinese’s government in offering test permits. These companies include WeRide.ai, Pony.ai, and AutoX.ai, some of them already running Robotaxi operations in the city of Guangzhou and Shanghai.
The execution system of an autonomous vehicle requires wire control technology. Taking a fuel vehicle as an example, the wire control technology mainly has five parts, namely, steer-by-wire, steer-by-wire, shift-by-wire, throttle-by-wire, and suspension-by-wire.
At present, European, American and Japanese suppliers have absolute advantages in the field of autonomous driving vehicle execution systems. The relevant wire control patented technology and component products are basically controlled in Bosch (Germany), China (Germany), Denso (Japan), Delphi (USA), etc. In the hands of a few large parts suppliers. Domestic technical reserves in China are relatively weak. China Jingxi Heavy Industry, Wanxiang Group, Wuhan Yuanfeng, Bethel, Yi Lida and other companies are accelerating their deployment. However, the above-mentioned parts giants are mostly leading in price and performance. As a results, the new Chinese companies need an accelerated plan in order to play catch-up.
R & D environment
In terms of the R&D environment, the development in China has shown much stronger support from the government. The strong management in China is mainly reflected in the standards for the road test of enterprise self-driving vehicles, while strong support is provided in terms of supporting infrastructure, as mentioned above.
For example, in order to meet the road test of domestic enterprises’ self-driving cars, relevant state departments have set up a test demonstration area for intelligent networked vehicles, including closed test areas and open test roads.
Among them, the closed test zone includes the state-level test demonstration zone supported by the Ministry of Industry and Information Technology and other ministries and commissions, such as the closed test zone of the National Intelligent Connected Vehicle (Shanghai) Pilot Demonstration Zone. As for the construction of open test roads, Beijing, Shanghai, Tianjin, Chongqing, Guangzhou, Wuhan, Changchun, Hangzhou and other cities have issued road test management specifications, delineating road test open areas, and successively adopting “5G + Beidou” vehicles Road collaboration network.
What’s more important is that the Hangzhou-Shaoxing-Ningbo Smart Expressway has started construction, with a total length of 174 kilometers and a total investment of 70.7 billion yuan. It is the world’s first smart expressway constructed in full accordance with the requirements of autonomous driving technology.
It is reported that there are mainly 4 application scenarios for the Hangzhou-Shaoyong Smart Expressway:
1. Accompanying information corridor. This is a customized push of traffic service information. Customize personalized traffic information services according to real-time events, vehicle locations and terminal types, and achieve precise push of guidance and travel assistance information for individual vehicles.
2. Active lane-level control. The six lanes of the Hangzhou-Shao-Ning Expressway will allow cars of different speeds to drive in different lanes in the future, reducing safety and driving efficiency problems caused by the difference in speed of vehicles in the same lane. Especially in long downhill and curve accident-prone areas, active early warning and control of vehicles.
3. Automatic rescue orders. When a traffic accident occurs on the expressway, the smart platform can accurately locate the accident vehicle through high-precision maps, realize automatic alarms, and automatically dispatch rescue forces.
4. Intelligent guidance and early warning for harsh environments. Some sections are located in mountainous areas, where there are many rainy and foggy weather. Through intelligent perception and integration with road traffic conditions, the smooth and safe operation of roads all-weather can be ensured.
In addition to the Hangzhou-Shaoxing-Ningbo Smart Expressway, there are also several domestic smart highways under planning and design in China. It is foreseeable that all our highways will be upgraded to smart highways in the future.
On the other hand, the Detroit-Ann Arbor corridor project has also been announced in August this year in the U.S. The Michigan State announced the project, which Dearborn-based Ford Motor Co. is renovating for $350 million. Cavnue, a subsidiary of New York City-based Sidewalk Infrastructure Partners LLC, was chosen by the state through a bidding process to lead the effort. From the project simulation, a vision for a multi-lane expressway accommodating both autonomous and non-autonomous vehicles will be created. Passengers on autonomous vehicles can also seamlessly switch between a “driving” mode and an “autonomous” mode, depending on their actual situations.