Liu Hong First published in Zhiche Technology .
PSD power system design "Power System Design" provides related technologies for power, automobiles, energy, transportation, etc.; Electronic Magazine covers leading domestic and foreign technologies, application cases and product information. Provide articles, news translations and soft article writing, and manufacturers are welcome to contact us. Free promotion: personal papers, marketing promotion, product release. 135 original content
/ Introduction Read /
As of September, Beijing, Chongqing, Wuhan, Shenzhen, Guangzhou, Changsha and other cities across the country have allowed self-driving cars to be commercially tested in specific areas and periods. Autonomous driving is moving from R&D to mass production, gradually entering the service operation stage, and the demand for cloud computing power is also rising, and "car-cloud integration" is on the verge of on the verge of being on the verge of being on the verge of being on the verge of being on the verge of being on the verge of being on the verge of being on the verge of being on the verge of being on the verge of being on the verge of being on the
Tencent Car Map Cloud Solution: Terminal + Cloud Integrated Map
Tencent Group Vice President and President of Tencent Smart Transportation and Travel Zhong Xiangping said: "Since entering the automotive industry, Tencent has always been user-centered, giving full play to technical capabilities, ecological connections and usage The three core advantages of household service are: the most advanced technical product capabilities in various scenarios through open cooperation, helping automobile companies to achieve the best service for users throughout the cycle. "
The car map cloud solution released this time relies on the safe and efficient Tencent Smart Car Cloud to provide the map update services and location services necessary for autonomous driving in an open mode, providing the cornerstone for autonomous driving service operation . ”
Tencent Group Vice President and President of Tencent Smart Transportation and Travel Zhong Xiangping
Tencent Smart Travel Liu Shuquan said: "As autonomous driving becomes a service, the iterative efficiency of autonomous driving functions has become the most critical indicator. Tencent provides open map capabilities with "car + cloud integration", and helps car enterprises build a large closed loop of "function release" and "service operation" on the terminal, driving the rapid iteration and experience upgrade of autonomous driving functions, successfully entering the mass production and operation stage . "
Call on the eve of mass production
As autonomous driving comes to the eve of mass production, it has gradually entered a generalized scenario, and bicycle intelligence continues to evolve towards "car-cloud integration". With the increasing demands of car companies for cloud computing power, the competition for cloud computing power has become a hot topic . What kind of cloud needs do behind the cloud-based trend of car companies? Why do we need to build a technical architecture and model of "car-cloud integration"?
Review of the automobile industry in 2022, under the trend of software-defined automobiles and "car-cloud integration", data and cloud computing have become key production factors and competitive factors in the automotive industry. Cloud has become a necessity, and automotive cloud has become a must-fight place in the industry. Top cloud service providers have launched smart car cloud-related products and strategies, and car companies are also building their own supercomputer centers to create automotive cloud strategic plans, and the cooperation model of car companies + cloud service providers has developed in depth.
Why does automotive cloud become popular? Public information shows that NIO ET7 can generate 8G per second after turning on intelligent driving. B's data is equivalent to watching 5-6 2K movies in 1 second, which is just the amount of data generated by a smart car. The
IDCh report predicts that the assembly rate of China's intelligent connected system in the automotive industry will reach 83% in 2025, with shipments to increase to 24.9 million units, and the global shipment of L1-L5-level autonomous driving cars in 2024 is expected to reach 54.25 million units.This means that with the continuous increase in the penetration rate of autonomous driving and intelligent connected and the improvement of automobile autonomous driving capabilities, the challenge of "data tsunami" has surfaced; cloud access has become a necessity for car companies to achieve more efficient and low-cost data storage, computing and model training, and accelerate product research and development and iteration.
cloud access helps car companies open up data channels in various business links such as R&D, production, sales, operations, and after-sales service, optimize business processes, and achieve cost reduction and efficiency improvement. Therefore, realizing the data-driven development of "vehicle and cloud integration" has become the key to the success or failure of the competition.
Since automotive cloud is a urgent need, how should car companies decide?
In fact, the current automotive cloud market is still in its early stages, and the demand for cloud services and cloud computing of car companies is also different. In the early stages of the development of , the primary demand of car companies is to "run" first, and when intelligent connected vehicles are mass-produced in large-scale production, only by striving for excellence can we form competitive barriers.
Tesla In August 2021, the self-developed cloud supercomputing center Dojo is mainly used to process and train AI software data of more than 1 million self-driving cars. Do domestic auto companies also want to follow the example of leaders with core competitive advantages in self-developed technology and build their own cloud service platforms? The answer is debatable.
Faced with the demand for large-scale mass production of autonomous vehicles, every car company has built an autonomous driving operation platform, which is indeed a bit difficult. After all, not only does self-built cloud infrastructure have huge investment in the early stage, but the subsequent results may not be satisfactory.
There are many roads leading to Rome, and there is not only one choice for car companies. Choosing a powerful third-party cloud computing company is becoming the ultimate choice for most car companies.
is self-evident. This cloud service manufacturer needs to have mature industry- proven experience, security compliance and risk controllability, which can help car companies quickly establish cloud service platforms, quickly respond to their next-generation model research and development needs, create differentiated products, and extend application scenario functions. At the same time, it must be able to ensure the security of car companies' data assets and help car companies to develop sustainably.
Intelligent driving cannot be separated from "car-cloud integration"
Vice President of Tencent Smart Travel and head of Tencent Smart Travel said in an interview: " cloud is becoming an important productivity, and the data driving of 'car-cloud integration' will become the key to competition in the automotive industry . In the future, intelligence, experience and new business models will be defined based on the cloud." Even if it is data-driven, "this data is the data of car companies' own data. We will not cross this field, but provide basic technical capabilities and services, rather than Tencent's operation and service methods." He said.
He believes: "' car cloud integration' is definitely a necessary infrastructure . From a global perspective, there is no autonomous driving just a bicycle runs directly. Its operation cannot be separated from the cloud, whether it is the mass production capacity or the ability in R&D."
He divides the foundation into two dimensions. One is the R&D demand . Bicycle intelligence is already There is no problem with running under trained models and algorithms, but algorithm training and continuous optimization cannot be separated from the cloud; is the second one at the operation level , which needs to be combined with the Chetu cloud solution, because today is not full-scene autonomous driving. For example, ODD (operation design domain) is just a specially designed operating conditions for autonomous driving and related functions, and is open on limited roads. When autonomous driving becomes a long-term service for users, how to evaluate autonomous driving functions in real time for users’ personalized needs and dynamically changing road scenarios, and iterative update algorithm, cloud capacity building is indispensable.
It has its own reason without its own car
In June this year, Tencent took the lead in proposing the "car-cloud integration" strategy . released a one-stop cloud solution "Tencent Smart Car Cloud" customized for the smart car industry and the "Tencent Integrated Automobile Safety Solution" covering the cloud pipe end in all aspects, and opened the industry's first cloud zone specifically for autonomous driving and smart cars in Shanghai, helping car companies build a "car-cloud integration" data-driven closed loop based on different scenarios, helping auto companies build a "car-cloud integration" data-driven closed loop based on different scenarios, helping autonomous driving and intelligent cars develop.
Regarding the open , Zhong Xuedan explained using autonomous driving as an example. On the one hand, the car-side algorithm, perception, decision-making and execution capabilities need to combine the actual execution effects through the cloud training platform, and continuously tune the algorithms to make it more valuable, so as to continuously iterate and upgrade the driving experience. "Of course, the closed loop of 'car-cloud integration' cannot be built at once, and it takes a long process. Our entry method is to continuously improve according to the actual needs and scenarios of customers."
He introduced that more than 100 car companies and mainstream travel technology companies have chosen Tencent Smart Travel' cloud services, and more than 40 car companies use Tencent smart cockpit products. For example, Tencent helped BMW build a digital ecosystem with Chinese characteristics. The two parties jointly introduced WeChat into BMW's car-end system based on Linux system, creating an in-vehicle light application ecosystem - a in-vehicle small scenario platform. It has now launched more than 30 high-quality in-vehicle small scenarios to provide services to more than 700,000 BMW owners.
In the field of digital marketing , Tencent helped FAW-Volkswagen build a private domain direct sales model covering the entire life cycle of users. With the support of Tencent's growth intelligent solution, FAW-Volkswagen's private domain user pool has expanded by 46%, the monthly active users of private domains have increased by 151%, the private domain clue conversion rate has reached more than 55%, and the cultural and creative mall profit has exceeded 70 million.
Recently, Tencent has deepened the strategic cooperation between NIO . Tencent will provide capabilities such as intelligent driving maps and car map cloud to help NIO create an innovative experience of driving together with standard navigation and high-precision navigation, and explore a scenario-based intelligent service network based on location services and Internet ecology. For advanced autonomous driving, Tencent's cloud services are also helping leading companies such as Mercedes-Benz , BMW, Bosch .
In terms of ecological opening , Tencent has introduced more than 600 industry partners to create joint innovation solutions to jointly provide better services to the automotive industry.
Not building cars is Tencent’s established strategy. Its foothold is the complete layout of the car end and cloud. Through cooperation, it helps car companies build and sell good cars, improve corporate management efficiency, and improve user travel experience. In smart car cloud, Tencent has expanded new solutions around the emerging new needs from the development to the operation stage of autonomous driving, using C2B capabilities and cutting-edge technologies such as cloud computing, AI, and big data to provide the automotive industry with a low-threshold customized platform and tool chain to create new productivity of cloud smart cars.
The problem is that as a company focusing on smart travel and autonomous driving, how can it turn its technological advantages into commercial value without its own car? Zhong Xuedan responded that Tencent is positioning as a digital assistant, helping auto companies and technology companies achieve better intelligent capabilities based on their car platforms. "We focus on three aspects: First, the ability to enhance digital basic capabilities with cloud map as the core; second, the ability to serve users, and third, the ability to connect with their core competitiveness, to help partners cooperate with their core competitiveness, complete, do well, and do things that they are not good at."
He said that today's cars have changed from individuals to an network node , and a strong network platform is needed behind them. This demand makes "car-cloud integration" and data drive the core capabilities in the key industrial competition. It is this ability that can better optimize vehicle-side perception and experience based on data, and use the data-based driver platform built on the cloud for continuous upgrade and iteration.
"All this is based on our digital assistant capabilities. From the core scenarios of building good cars, using good cars and selling good cars, it helps car companies establish a closed loop of vehicle-cloud integration and data-driven data-driven around factories, stores, people and cars, and improve the production efficiency of each link. We are constantly improving data and computing capabilities construction, and do not consider building a corresponding hardware platform." He said.
Since Tencent’s “car-cloud integration” provides data-driven infrastructure for the automotive industry with cloud as the core, how to solve the difficult problems in the automotive industry in a targeted manner?
Zhong Xuedan said that Tencent infrastructure consists of "three ends" - mobile phones, cars and cloud, and seamlessly connects to achieve the organic integration of scenarios .
On the mobile terminal , Tencent has the basic advantages of WeChat + enterprise WeChat connection capabilities. It can provide good service capabilities through WeChat's "Tencent Travel Service" mini program and other products to achieve seamless connection between mobile phones and car scenarios. The rich in-car entertainment ecosystem and content can allow users to experience a perfect experience in the interior space that is different from other scenes.
On the vehicle side , Tencent's smart cockpit TAI, smart driving maps, high-precision maps, smart scene engines, small-car scenes and other products and capabilities can all help the evolution of intelligence on the vehicle side. For example, for future intelligent navigation for people and cars, use standard maps, ADAS maps, and HD maps to achieve smooth switching of different driving modes such as manual driving, assisted driving, and intelligent driving. For example, the commuting scenario, combined with "car-cloud integration", through continuous learning and training, vehicles can understand some special scenario needs on the daily road, continuously reduce driving difficulties, and improve the autonomous driving experience; in urban market scenarios with very high difficulty, more accurate data is used to feed back the changes in road signs to users, and optimize path planning.
Zhong Xuedan emphasized: "Only by combining mobile phone, vehicle and cloud, better and more appropriate service capabilities can be provided according to user specific needs, helping users establish a data-driven closed loop through effective digital connections, continuously update and iterate, improve the efficiency of each link, reduce costs and increase efficiency."
In order to solve the difficulties in fusion, Tencent has established a better identification between user ID, car and mobile phone, and opened up mobile phone car connection and service channels, especially to ensure data consistency ; at the same time, combined with the characteristics of the car, it is necessary to enhance services, such as continuously providing service capabilities that meet the needs of in-car entertainment, conferences and other scenarios.
Regarding Tencent Map's advantages , Liu Shuquan gave three key points. First, the data is relatively fresh, and the collection system and data system have been designed for intelligent driving from the beginning; second, the data integration of three maps (SD standard map, ADAS lane-level map and HD high-precision map) data is the same source, with better integrity and consistency; third, a complete and open closed-loop system can target the partners within a safe and controllable framework, from data services to engine capabilities, and then to the upper-level customized services functions. Based on these three points, Tencent Map can share various services with customers.
Liu Shuquan, Vice President of Tencent Smart Travel,
"Compared with other maps, we have the most accurate POI information, because applications such as WeChat and Meituan are all implemented using the basic positioning data of Tencent Maps." He added.
Smart Car Cloud "Special Cloud Dedicated"
It is not difficult to imagine that the more efficient and low-cost way of "going to the cloud" must be "Special Cloud Dedicated", but the industry cloud that meets the specific needs of the industry.
As the industry's first one-stop solution specially designed for smart cars, Tencent Smart Car Cloud organically integrates all-round capabilities such as autonomous driving R&D and operation tool chain, smart cockpit R&D, smart driving maps, and operation services. Through the "one cloud, one-stop" method, it meets the wide range of smart cars' needs from R&D to operation, helps customers reduce the complexity of architecture and obtain one-stop optimal solutions.
Tencent opens the East China Zone of Smart Car Cloud in Shanghai. It is the industry's first safe and trustworthy proprietary cloud platform built specifically for the fields of autonomous driving and smart cars. It is understood that at present, Bosch, Mercedes-Benz and other car companies have used this platform to train autonomous driving algorithms, improve algorithm capabilities, and optimize tool chain scenarios.
is one aspect of meeting specific needs of the industry, and more importantly, cloud compliance, and scalability . According to Liu Shuquan, Tencent Cloud 's cloud management platform CDZ can provide customers with value-added management services. Based on Tencent Cloud's homologous base, Tencent Smart Car Cloud can use products with open services on Tencent Cloud to seamlessly expand into the proprietary cloud of Smart Car Cloud according to customer needs. Its scalability has two dimensions: one is the product dimension, and some on the public cloud can be deployed; the other is regional scalability. The CDZ software combines the backbone network to make services into two places, three centers, multiple places and multiple centers, and other methods to expand horizontally. The deployment method, location, and capacity are very flexible.
Based on the cloud zone, Tencent has created a complete cloud automation tool chain such as autonomous driving, simulation training, high-precision maps, cockpits, etc., providing car companies with proprietary cloud services out of the box, allowing car companies to focus more on algorithm optimization and experience improvement . Smart car cloud is physically isolated on the full stack to ensure the security and trustworthiness of related data and the independent and controllable of . At the same time, cloud components will be targeted to optimize the special needs of the industry, truly achieving "special cloud and dedicated use".
For autonomous driving data processing , Tencent Smart Car Cloud provides GooseFS storage acceleration services, supports access efficiency of 100W QPS under trillion+ file scale, and ensures low latency. Compared with traditional storage access and access modes, its acceleration performance is 10 times higher; for data computing and model training, one-stop algorithm development and training framework - TI-One can save a lot of algorithm training costs. In the data access process, up to 80% of the workload can be saved; in the data processing process, 70% of the labeling cost can be reduced; in the model training process, algorithm development TCO can be reduced by at least 50%.
Car map cloud solution - the cornerstone of autonomous driving service operation
Judging from the situation in 2022, the penetration rate of advanced assisted driving is increasing rapidly, becoming an absolute industry hotspot. The access efficiency, computing efficiency, storage costs and service releases of the cloud behind it are increasing, and the first year of "vehicle and cloud integration" has arrived.
When the mass production of autonomous driving is implemented, Generalized scenarios - from test areas to ordinary roads, from highways to urban roads, autonomous driving has also evolved into a continuous user service, and users have put forward more demands on the safety, comfort and humanization of autonomous driving. In this context, an autonomous driving operation platform has become a necessity. To meet the needs, professional operation personnel must be responsible for the safety and humanized service experience of autonomous driving. It also must have safe, stable and cloud-based operation tools. Based on accurate and rich map services, it can record, evaluate, iterate and upgrade autonomous driving capabilities at any time to achieve the operation of autonomous driving service throughout the life cycle.
Tencent’s integrated car map cloud solution combines map and cloud capabilities, reflecting the following core values:
Customizable map update service
The second is the autonomous driving operation location service . Through extensive and vivid roads, POIs, traffic information, etc., it helps operators comprehensively evaluate autonomous driving performance; combine user trajectory, hot sections, vehicle data, etc. to conduct user value analysis, model optimization and road tests, and perform orderly function upgrades, accelerate the iteration of autonomous driving functions, and improve user experience. In terms of autonomous driving road operation, by analyzing the operating performance on specific roads and comparing them with benchmark data, functional optimization is carried out; hot sections and special structural roads are mined through maps, and data collection, directional road testing, and directional optimization of the autonomous driving functions in iterative operation.
Automatic driving operation location service
At present, domestic OEMs are no longer satisfied with navigation data and positioning capabilities, and need precise map services, and cloud is the best way to implement continuous services. Liu Shuquan believes: "Only a platform with fresh data, flexible iteration, and better definition of data APIs can bring the greatest value to customers. We can provide one-stop service, and customers also choose cloud service , simulation platform or other tool chains according to their needs, and partners can also put the tool chain they developed in."
and the "most knowledgeable" partners are in the same way as
Professional things need to be handed over to professionals. To seek development, we need to rely on the "most knowledgeable" partners to move forward side by side.
Liu Shuquan said, "To realize the digitalization of the smart car industry, the trust relationship with partners is particularly critical. . The closed loop of end-cloud integration is not something Tencent can accomplish, but requires continuous optimization with customers in every link."
Zhong Xuedan finally said: "Multi-cloud coexistence is the future trend. To meet different scenarios, whether it is public cloud, private cloud or proprietary cloud, the automotive industry will have corresponding needs. As a digital supporter of the automotive industry, Tencent will continue to deepen its digital infrastructure with cloud map as the core and launch solutions that highly match industry needs; at the same time, it will achieve joint innovation with a more open ecosystem and a more 'knowing' partners in all links of the automotive industry."
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-