Baker Street Detective Officer
Author: Lu Zhenxi
Putting aside security and talking about commercialization, it is either rogue or nonsense
In October 2021, Beijing allowed driverless road test drivers without safety officers. This means that driverless cars are one step closer to the commercial operation stage. As of January 2022, 8 provinces, including Shanghai, Anhui and Jiangsu, have successively issued test management rules and subsidy support documents.
As can be seen from the above figure, the provincial rules and supporting documents are based on the "Management Specification for Road Testing and Demonstration Application of Intelligent Connected Vehicles (Trial)", which is the basis for the commercialization of automatic driving.
The management specification document clarifies the management norms that all parties involved in the test should comply with from six aspects: road test and demonstration application subjects, drivers and vehicles, road test applications, demonstration application applications, road test and demonstration application management, traffic violations and accident handling, and also clears the policy obstacles for autonomous driving enterprises to carry out trial operations.
Autonomous driving enterprises combing
In fact, as early as the release of the specification document in July 2021, Baidu Apollo has already launched the normal operation of Robotaxi, and it is also the earliest enterprise in China to enter Robotaxi.
Baidu Apollo officially launched its normal operation activities in Robotaxi in May 2021. In August 2021, Baidu released a new and upgraded self-driving travel service platform - "Radish Run". At present, the cities operated by Radish Run are Beijing, Shanghai, Changsha, Guangzhou and Cangzhou.
According to public information, the unit cost of Baidu Radish Run's model Apollo Moon is 480,000 yuan, about one-third of the industry average. Baidu expects that in 2023, the operating cities of Radish Run will be expanded to 30, serving more than 3 million users and orders in the millions.
In addition to Baidu, there are also a number of domestic enterprises to lay out the Robotaxi market. From 2016 to 2020, Didi, Mushroom Car Union, CIIC, SAIC Motor, GUANGZHOU AUTOMOBILE GROUP, Dongfeng Group, etc. have successively entered the game.
Didi
The first is Didi, although it started late, but relying on rich platform experience, it quickly promoted the testing process. Didi Autonomous Driving started in 2016, launched a passenger pilot in Shanghai in June 2020, and obtained the first batch of manned demonstration application licenses in Shanghai in September.
In the first half of 2021, Didi reached strategic cooperation with GAC Aean and Volvo on autonomous driving. As of June 2021, Didi has more than 500 autonomous driving team members and more than 100 self-driving cars, and has obtained road test qualifications in Beijing, Shanghai, Suzhou, California and other places, with a test mileage of more than 530 kilometers in Shanghai.
Mushroom Car Union
In 2017, by cutting into the field of urban public travel and public services, Mushroom Car Union provided operational services with a vehicle-road-cloud integrated full-stack solution to achieve a new breakthrough in the large-scale landing of Robotaxi business.
The company independently developed the automatic driving system of vehicle-road-cloud integration, landed the first open 5G commercial vehicle road collaborative demonstration road in China in Shunyi, Beijing in 2019, and signed a strategic cooperation agreement with the Hengyang Municipal Government in 2021, running through the main roads of Hengyang City, with a total mileage of 200 kilometers.
Through the combination of government and enterprises, Mushroom Car Union improves the resource reuse rate and scale effect, gives full play to the advantages of the integrated solution, and solves the high cost problem of Robotaxi. At this stage, Mushroom Car Union has landed autonomous driving commercial projects in Beijing, Shanghai, Jiangsu, Hubei, Hunan, Henan, Yunnan, etc., with a total order volume of more than 2 billion yuan.
In 2018, CIIC achieved the overall capability of the system beyond L4 level automatic driving by creating advanced intelligent road (L4) and general vehicle intelligent (L2.9) technologies, and promoted L4 cost reduction and efficiency improvement and large-scale landing.
CIIC
In October 2021, CIIC successfully demonstrated "road driving" on the highly intelligent road in Xiangcheng District, Suzhou City, including three scenarios such as "digital road closure, multi-vehicle collaborative U-turn, and blind spot car concession", and received hundreds of people for a test ride experience. CIIC is currently conducting technical verification in Suzhou, Shanghai and other places.
SAIC Motor
In addition, SAIC Motor also cooperated with Momenta to launch the robotaxi trial operation in Shanghai, and in December 2021, SAIC Motor announced the adoption of Momenta L4 level autonomous driving solutions, and launched 20 operating vehicles in Shanghai and Suzhou, and is expected to land in Shenzhen in 2022 and achieve an overall operating fleet size of 200 units.
GAC Group
GAC Group joined hands with Wenyuan Zhixing in Guangzhou to start the trial operation of Robotaxi. According to public information, Wenyuan Zhixing has more than two years of Robotaxi operation experience, and the driving road test mileage exceeds 8 million kilometers.
In August 2021, GAC was awarded the Guangzhou Road Test Test License, and the fleet can carry out autonomous driving demonstration operations in Panyu Automobile City, and three months later, GAC Group, Wenyuan Zhixing, and Ruqi Chuxing announced a strategic cooperation to jointly promote the research and development and operation of Robotaxi, and plans to launch Robotaxi operation on the Ruqi Travel platform in 2022.
Dongfeng Group
Dongfeng Group joined hands with Yishi Technology to open a trial operation of Robotaxi in Wuhan. In September 2019, Dongfeng obtained the first self-driving car road test license in Wuhan. In February 2021, Dongfeng Autonomous Driving Robotaxi was officially launched, which adopted the L4 level of autonomous driving solution of Yishi Technology, and the launch of 42 vehicles was mainly located in the Wuhan Economic Development Zone, with 22 stops. It is expected that more than 200 autonomous vehicles will be launched in the next two to three years.
Pony Zhixing
Finally, the company has joined hands with Nvidia to promote a new autonomous driving system and launch a self-developed vehicle specification-level calculation unit program, equipped with Nvidia DRIVE Orin? (SoC) system-on-chip is driving mass production of the next generation of autonomous driving hardware and software systems to accelerate the large-scale deployment of L4 level autonomous driving technology, and the new computing unit will begin mass production by the end of 2022.
At the same time, Xiaoma Zhixing released the sixth generation of automatic driving system solutions, the first model equipped with the system is Toyota "Xena SIENNA" Autono-MaaS vehicle, which will start road testing in China this year and put into the daily operation of autonomous driving services (Robotaxi) in the first half of 2023.
However, it is worth noting that the progress of the self-driving project of Xiaoma Zhixing is not smooth, according to the news on March 9, Xiaoma Zhixing recalled three test vehicles in the United States due to automatic driving system problems, due to a lack of safety personnel, one of which hit the lane divider and road sign in Fremont.
Steven Cliff, director of the National Highway Traffic and Safety Administration (NHTSA), said in a statement in response to the recall that whether the vehicle is operated by a human driver or an autonomous driving system, the agency always puts the interests of road users first.
In addition, according to Reuters, on March 7, local time, Intel's self-driving car subsidiary Mobileye submitted an initial public offering (IPO) application for listing on the US stock exchange, which is expected to become one of the largest IPOs of US stocks this year. According to people familiar with the matter, Mobileye's valuation at the time of its listing will exceed $50 billion, far beyond Intel's valuation of about $15 billion when it made the acquisition in 2017.
From the perspective of project progress, Baidu benefits from the first-mover advantage, which is the largest fleet among the existing layout enterprises, but is the layout really equal to the scale? For any emerging project, solving the pain points of users is the only way to continue its sustainable development.
The premise of solving user pain points is security
At this stage, the biggest pain point of user travel, one is the road congestion delay time, and the other is that the car is not good enough to dare to drive. Baidu Apollo, which is currently operating, can uproot the above two pain points.
The first is the problem of road congestion, Apollo usually does not need a safety officer to take the helm, and in special cases, the cloud driving function can be achieved through 5G, and the operation of the automatic driving platform will continue to be handed over to the automatic driving platform after passing through the special road section, and the user can greatly use the ride time to handle personal affairs. It is precisely because of the characteristics of "driverless" that users with general driving skills can also enjoy a pleasant travel experience, thereby expanding their user scale.
Since the realization of autonomous driving open operation in Beijing in 2020, Baidu Apollo has gradually launched unmanned testing, and has accumulated 20 million kilometers of test mileage by 2021. It should be noted that the complexity of domestic urban road scenes is much higher than that of mainstream road scenes in Europe and the United States, and the innate "learning attributes" of unmanned driving systems can continuously iterate themselves in the service process.
For unmanned driving, it eliminates the external manifestation of the driver, and expands the market in addition to solving user pain points, but also convinces users to believe that it is safe and reliable. With the main system, redundant system and cloud driving three-layer security system, Apollo brings users an acceptable safe driving experience.
The first is the basic OS within its main system, Baidu's self-developed automatic driving system has been open source for many years, compared with the industry system, Apollo has the characteristics of high stability, high real-time and low latency, which is currently widely accepted by the industry, the system combines algorithms, so that the system reaction speed and delay are better than human drivers.
The second is the pan-perception system, which includes three aspects: map, positioning and environmental perception, to achieve accurate portrayal of environmental scenes. Map naturally needless to say, Apollo's positioning system is based on the fusion scheme of multi-sensor positioning, which can achieve stable and accurate positioning under working conditions such as signal occlusion and environmental changes.
The perception system is based on strong visual perception capabilities and has stronger capabilities for long-tail scenes. Baidu Apollo focuses on optimizing the detection capabilities of difficult scenes such as low-altitude occlusion, so that it can accurately provide stable environmental perception under various working conditions. Moreover, domestic urban road construction is more frequent, and the automatic driving system map update will inevitably be delayed in the actual road conditions, so Apollo will also match the on-board map with the on-board map by detecting the road environment in real time, and timely discover the changes generated by the environment compared to the on-board map to avoid the vehicle entering the dangerous scene.
The third line of defense to protect the safety of users is the decision-making planning system, which directly determines the driving behavior of the vehicle and is directly responsible for the safety of the vehicle. Apollo first abstracts out the complete strategic expectations in each scenario according to the current traffic regulations, and at the same time, for scenarios with certain potential risks such as occlusion, Apollo decided to adopt a defensive driving strategy to ensure that in the worst case, the vehicle can still minimize possible safety hazards and ensure safety and controllability.
Finally, to further improve driving safety is the vehicle-road collaboration system, which is an effective supplement to the intelligence of bicycles, and the existing vehicle-road collaboration technology based on roadside perception essentially provides perfect over-the-horizon perception capabilities. Based on this, the automatic driving system can realize the early perception of typical dangerous scenes such as occlusion, and the vehicle-road coordination technology has the ability to observe and stabilize the viewing angle for a long time for low obstacles, so as to achieve more accurate identification and assist the autonomous vehicle to make more reasonable avoidance behavior.
The second layer of protection is the monitoring and redundancy system, the main purpose is to find the safety risks in time when there is a safety risk in the software and hardware of the vehicle itself, or the scene in which the vehicle is located, and to keep the vehicle in a safe state, which is also a necessary condition for achieving unmanned driving. The last layer is cloud driving, which transmits the car's video back to the cloud in real time, and the security officer in the cloud understands the vehicle dynamics in real time and controls it in real time.
Through the interoperability of these three-tier systems, Apollo can already provide users with an acceptable driving experience, further increasing user viscosity, scaling up, and laying the foundation for full commercial applications. From the application level, Apollo's driverless platform is close to the car.
Image source: Apollo3·8 press conference
More importantly, after Apollo opened commercialization, the user scale, frequency and stickiness showed steady growth, and the increase in new users in the first month of commercialization reached 151%, and the per capita number of new users in the first week and the proportion of people who repurchased more than 4 times have increased significantly. User satisfaction reached 4.8 points under the background of 5 points of satisfaction, which provided a strong support for the realization of the project.
Because no matter what the project, if you want to make a profit, one is used, and the other is that the satisfaction after use is higher, the former provides a stable basic disk for the project, and the latter brings unlimited growth streams to the project, after all, the gold cup and silver cup are not as good as the user's reputation. With a good user reputation, can ensure the longevity of the product, for the brand, it is more beneficial and harmless, when Apollo gradually increases the number of application cities, its future scale is immeasurable, will become Baidu's next revenue growth pole.
Lidar escorts autonomous driving
The commercialization process of driverless enterprises led by Baidu is gradually accelerating, which will not only bring new growth poles to the enterprise itself, but also open up new growth space for the entire autonomous driving industry chain.
In the next decade, the continuous improvement of the market share of new energy vehicles, as well as the popularity of private cars, will promote the development of the overall shared travel market; coupled with the increasing habit of consumers to share travel services, the overall crowd penetration rate will gradually increase in the future, participants and participation methods are also constantly changing, the development of intelligent automobiles and the popularization of autonomous driving technology will promote the penetration of driverless vehicles.
IHS Markit expects the market size of the entire shared mobility market to reach 2.25 trillion yuan in 2030, with a compound growth rate of between 20% and 28%. In the shared mobility market, Robotaxi will account for more than 60%, and the market size will exceed 1.3 trillion.
In this trillion blue ocean market, one of the more noteworthy market segments is vehicle-mounted lidar, which is the perception system in the autonomous driving safety system. From the perspective of the development process of automatic driving, multi-sensor fusion is already the trend of the times, and as far as the current technical means are concerned, lidar may be the final choice.
Because of the existing ADAS system, the main visual sensor still relies on the camera, although it is one of the most mature on-board sensors, but because the camera is the same as the human eye, it is passive to receive visible light, so the visual effect is poor in the case of backlight or complex light and shadow, and it is susceptible to bad weather.
Although millimeter-wave radar and lidar work similarly, and there are characteristics of long enough wavelength, good winding ability, and minimal influence by the weather environment, but due to the excessive wavelength, the detection accuracy is greatly reduced. This is also its biggest shortcoming and cannot be changed.
In order to make up for this shortcoming, lidar came into being, lidar uses laser as a carrier, the wavelength is shorter than millimeter wave, so the detection accuracy is high and the distance is long. Lidar can also form a 3D "point cloud" image of obstacles by recovering information from laser rulers in different directions and forming lines with lines and lines. Due to technical difficulties and high costs, large-scale loading has not been achieved before and has only been widely used in the Robotaxi market.
Compared with the passenger car field, Robotaxi has extremely high requirements for perceptual layer detection performance due to its positioning of L4-L5 level driverless. And because the owner of the vehicle, unmanned operation driving company is often a strong capital strength of the car / technology manufacturer, so the price of lidar and the integration with the body requirements are relatively low, a number of operators Robotaxi have been equipped with lidar, is the main battlefield of the current lidar.
However, with the continuous improvement of the level of autonomous driving of passenger cars, a large number of companies have emerged in the passenger car market, and the mass production model equipped with lidar has been launched, and the market share of lidar will be further improved in the future.
Looking at the domestic lidar supply chain, a large number of enterprises have emerged in China to participate in the manufacture of lidar, and with the continuous iteration of related enterprise products, the price of finished lidar is expected to be further reduced, thereby increasing the installed capacity of bicycle radar, further protecting the safety of vehicle driving, and improving the driving experience of users.
When the safety of automatic driving is further improved, the superimposed Apollo-based autonomous driving brand continues to exert efforts at the marketing end, the public's acceptance of automatic driving and unmanned driving will be further improved, and its market scale will also resonate with it.