Tencent Shi Xuesong: Creating a two-wheel digital base driven by AI to help transform the transportation industry

2023-09-12 14:53 0

On September 8, the 2023 Tencent Global Digital Ecology Conference on Smart Transportation was held in Shenzhen. Shi Xuesong, vice president of Tencent Intelligent Transportation, delivered a keynote speech entitled "Tencent Digital Base, Helping the transformation of the transportation industry".

At present, the development of transportation is shifting from scale growth to high-quality growth, and innovation in transportation science and technology has become the main driving force. As the level of digitization of transportation continues to increase, the industry is paying more attention to the deep application of digital technologies, focusing on intelligence to improve infrastructure efficiency, transportation safety and efficiency, and reduce costs.

Shi Xuesong, Vice President of Tencent Intelligent Transportation, head of Tencent Intelligent Transportation

Shi Xuesong pointed out that the future transportation life needs to build a loop from full perception to calculation, decision-making, and touch, and has the ability to evolve on its own. In this regard, Tencent proposed the two-wheel architecture driven by AI as a universal technical base for digital transportation.

Shi Xuesong said that real-time twins and transportation OS are two key parts of this base. Real-time twinning provides high quality and high value data basis for large AI models by recording dynamic and dynamic scene data of transportation. Tencent also makes full use of existing digital tools to break digital silos, carrying applications in the form of traffic OS and conducting automated control. Based on this base, the design architecture of intelligent transportation can be reconstructed.

Shi Xuesong stressed that the construction and iteration cycles of transportation infrastructure and digital facilities are not synchronized, so in the process of digital transformation, it is necessary to break the information island with a layered and decoupled design, so as to achieve rapid evolution at all levels. Tencent hopes to make full use of the technology and capabilities accumulated in the past two decades, adhere to the concept of "people-oriented" to develop intelligent transportation, implement product methodology, and bring safer, efficient and beautiful travel services to the people.

The following is the full text of Shi Xuesong's speech:

Good morning, everyone. I'm Shi Xuesong from Tencent Smart Transportation.

In China, the construction of transportation power has entered a new stage with high-quality development as the goal. High-quality development includes two aspects: high-quality operations and high-quality services for travelers. Today, I will share some of Tencent's insights and practices in smart transportation construction in the past few years.

In April this year, the Ministry of Transport, the National Railway Administration, the Civil Aviation Administration of China, the China Post Bureau and the National Railway Group jointly issued the Five-Year Action Plan to accelerate the construction of a transportation power, marking the construction of a transportation power has entered a new stage of integration of digital technology and the transportation industry. In 2020, Tencent proposed to build the future transportation life body, and in 2021, we proposed to achieve this goal with a digital base, and adopted the AI+ two-wheel drive strategy to build it. One wheel of the two-wheel drive is the real-time twin, and the other is the traffic operating system.

Tencent's real-time twin is to use the digital twin method to record the scene of the combination of traffic and movement. With the help of real-time twinning method, a large amount of valuable data can be accumulated in the transportation system, which will be an important basis for large-scale model processing. In the transportation industry, real-time twins can create valuable and high-quality data for us. We can also make full use of the existing digital tools to open up many islands formed in the process of information construction in the past, and present them in the way of transportation OS. With a large amount of data and an application environment set up, we can use AI tools to achieve a loop from full perception to calculation, decision-making, and access, and become a future traffic life that can evolve on its own.

Total perception is very important for transportation. When providing transportation services, we need to have a good perception of the objects and operators of transportation. Because transportation not only has a complex infrastructure, but more importantly, it carries mobility, logistics, people flow, and traffic flow. If you can describe the results of these perceptions in a semantic way, it is the best language, and it is very critical for us. A typical application of perception is the field of containers, and we can use visual processing and analytical processing to achieve intelligent box inspection in this small scene.

Calculation is very important. The core symbol of Tencent's digital base is computability, such as in the direction of traffic, how the Internet big data and the real-time data generated in the process of transportation operations to do integration and interaction application, Tencent has been exploring the direction.

Today we are moving towards a closed loop in this direction. For example, today we build a base with BIM+GIS, which can solve spatial computing problems. But in Tencent's real-time twin,BIM+GIS is only static, and we have to add time to it to become space-time computing.

After adding the time dimension, the traffic scene can be described and recorded, otherwise only the information can be recorded, and the scene can not be recorded. Therefore, BIM+GIS realizes spatial calculation, and after adding perception and time relationship, it is upgraded to four-dimensional space-time calculation. When this digital base is completed, it can generate a large amount of high-quality data including transportation scenarios.

In terms of traffic control and induced decision-making, after the previous digitization, we can use these high-quality data to make decision inference. Here, Tencent's simulation technology and big data analysis technology are used, and Tencent's traffic OS can also be automated control. Using these data to do traffic simulation and analysis, the convergence rate is very fast, there is an order of magnitude improvement. In this direction, after we use the real-time twin method to restore the traffic dynamic and static, we can regenerate many new scenes, and these new scenes combined with the current real-time data can have a good impact on traffic control. It can also be extended to more scenarios, combine more patterns, and extend good methods and means. Tencent also provides traffic OS, it is a tool, but also an environment, down to provide the Internet of things, up with a low code way, you can create traffic control applications through drag and drop operations. This allows ordinary operators to bridge the code gap to build traffic applications.

Reach is what matters today, how we deliver high-quality, traveler-oriented services. Tencent's philosophy is people-oriented, which means that every user who uses Tencent travel services on the road must feel that the services we provide are good.

The automotive industry today regards the car as a very important product, attaching great importance to the interaction between manufacturers and owners, and the construction of cockpit contacts; Our road is also a product, and it is important that we build it not just for people to use, but to build a long-term relationship with all the users who use it.

For example, the road is expanded to 16 lanes in both directions, so how to let the owners of the road get better guidance is very important. Today our navigation may say "turn right 250 meters ahead, don't go uphill", but many people have a hard time establishing the concept of 250 meters of space, and it is easy to miss the exit. But navigation can be much more accurate if it is reduced to three simple instructions: "Change lane right, change lane left, and do." But the important premise is that navigation requires knowing exactly where the vehicle is. With the real-time twin method, we have been able to provide accurate position and situational awareness to every car on the road, and only an ordinary communication terminal in the car can well present the relationship between people, cars and roads, so that complex induction can also be simple to achieve through it.

Therefore, when real-time twinning is achieved to a certain extent, we have a good platform architecture and a good tool, and the road operator can already allow the command and monitoring screen, field operators and end users to obtain homologous data, which is very important. If a traffic accident occurs in a lane of 16 lanes in both directions, it needs to be closed, induced, and the rescue vehicle can use this means. So today, through such a real-time twin platform, the road company can already make the service ability to do precise interaction with specific users, which is a very good resource. Then we can use the product methodology to build new designs.

After building these bases, we can continuously accumulate various scenarios, various control capabilities, and various service capabilities on them, and can constantly evolve automatically, which is a very important ability. After building an information platform, IT used to be maintained by IT engineers; Today's base is different from the past, in addition to routine, typical operation and maintenance work, more importantly, it can evolve on its own.

Finally, a brief summary. In the development process of intelligent transportation, we must see that the construction cycle and iteration cycle of transportation infrastructure are much longer than the cycle of information systems and digital systems, and they do not match. Therefore, in the process of construction, we must do hierarchical decoupling, and then we must do cross-sharing.

Today, Tencent uses distributed cloud at the bottom, cloud-native digital twins and transportation OS above, and then achieves a good layerization through the deployment of AI platforms, supporting the rapid evolution of different levels respectively. It can also continuously improve its capacity through its own accumulation. Many problems can be solved by a layered approach. Some people think that large systems should also integrate software and hardware, but it is easy to create islands. When it comes to the digital transformation of the transportation industry, this combination of hard and soft systems is an easy way to exploit. So today in the direction of traffic intelligence, if we can build the base well, and then adhere to the people-oriented, implement the methodology of the product, we can get a good harvest.

The above is my experience in the past few years to share with you, thank you.

Source: Corporate press release
Keywords: Tencent digital AI
Press release Overseas media release advertorials Release advertorials release press conference Release press release overseas media release media release platform media release release press release Invite media to invite overseas press release Overseas press release