After graduating, he worked on racing games for several years. After the company was acquired by Apple, he switched to autonomous driving. A few years later, he left Apple and became the technical director of Tencent’s digital twin simulation technology. Sun Chitian has a wonderful career change experience.

Several jobs seem to be completely different from each other, but in fact, they have a common technical point behind them, that is, game technology. As he said, a gamer is not just an expert in playing games, but also a developer of games, but also a person who uses game technology to develop cutting-edge technologies.

Sun Chitian, who has loved games since he was a child, fell in love with mathematical physics after he understood the truth that “games should not be eaten”. However, when he was studying mathematical physics at Tsinghua University, he found that he lacked a little talent for being a scientist, and he still loved games. . So he started to learn game development from a graduate student. After graduation, he went to Silicon Valley. After developing a very successful AR racing game, he joined Apple’s autopilot team, and then came to Tencent to work on an autopilot simulation system, and then developed this digital twin simulation technology. Apply to the larger industrial Internet field.

In his view, the autonomous driving simulation system is similar to a “high-precision racing game”. The current appearance of the metaverse will be interpreted by many people as “Player Number One”, a world where everyone is playing games. In fact, that world will indeed be filled with games, but not just entertainment games, but also autonomous traffic, construction, and factories. The “gamification” of all walks of life, such as manufacturing, is the world of digital twins.

As the virtual and real worlds become more and more inseparable, behind Sun Chitian’s work experience, a revolution in “games” is taking place-game technology is not only for entertainment, but has become an important productivity tool.

The following is the speech transcript of Sun Chitian, director of Tencent Digital Twin Simulation Technology, at the Geek Park Innovation Conference IF 2023, organized by Geek Park:

01

from racing games,

to digital simulation

Xiaobawang game console, Gameboy, PS2, etc. are the game consoles I have played since I was a child. I am a person who loves games very much. It can be said that my entire growth process is related to games.

I was interested in mathematical physics when I was in high school, and then I went to Tsinghua University to major in mathematical physics. I found that the real systematic mathematical physics is very different from middle school. So I found that talent is very important for learning mathematics and physics. I can only be regarded as a hobbyist. It is too difficult to be a scientist. So later I learned computer graphics. The computer courses in my undergraduate period also involved simple immersive experience and game development, which laid the foundation for me to make games later.

I love playing games so much that I even applied directly for a game development degree in graduate school, where the courses were mainly on application development of games and immersive experiences using game engines. It was 2012, and game engine technology was not used much in China. For me, a person with a background in mathematics and physics, it took a lot of time to learn and understand.

During my study, I used the game engine to do a lot of game development. There are many types, whether it is ordinary games or AR , VR and other interactive and immersive games.

After graduation, I went to an AR development company. It was 2014, and AR was very hot in Silicon Valley. Many companies were exploring the use of AR to develop games, and we were one of them. At that time, we used Intel’s Real Sense camera to help develop AR immersive games. Users can control game operations in AR through the camera. It is a very useful AR interactive device. It was a racing Lego IP game. At that time, the game, interactivity, and AR experience of this product were very strong and very popular.

After making two games, this AR company was acquired by Apple, and I joined Apple. At that time, I had two career development directions to choose from, one was AR glasses and the other was self-driving cars. I was very confused at the time, and my manager told me that it would be cooler to use your background and previous game development capabilities to develop autonomous driving, especially the simulation system required for developing autonomous vehicles, which will have a great impact on the industry. good help. I also think autonomous driving is cool, so I joined the autonomous driving team to help the development of autonomous driving by building an autonomous driving simulation system.

So what is the simulation system needed for self-driving cars? You can understand it as a high-precision racing game. By creating a highly realistic environment, we let the self-driving car conduct a complete scene test in the virtual environment. We think that the self-driving car is tested very well in the simulated environment and runs very smooth and stable. , we will test self-driving cars in the field.

After returning to China, I joined Tencent, and I started to do autonomous driving simulation related work. We have redesigned and built an autonomous driving simulation system product with Tencent characteristics, not only for our own use, but also for domestic and foreign automakers, testing grounds, and national ministries and commissions. It is a domestically controlled autonomous driving simulation system.

The route I chose at the time was game technology plus data-driven. Data-driven is to collect a large amount of real data, such as real car data, dynamic data, and sensor data. These collected real data train and calibrate the automatic driving simulation system.

Game technology is also a very core part of the technical route. Tencent has a solid game foundation in the past few years, such as game engines, data synchronization, and scene generation tools. Take game engines as an example, such as how a camera and lidar work and operate, and how to make the simulation results and real ones The result is close. It is necessary to simulate the process of lidar, reflection, generate a complete point cloud, and automatically construct and render the scene. The whole process needs to be realized with the help of the game engine.

The technical route at that time was very sharply challenged and criticized by traditional simulation software, saying that “the use of the game engine is not rigorous enough.” But in fact, five years have passed, and the result is that all autonomous driving simulation software supports game engines. The advancement of game technology can help us save a lot of time in the simulation industry, solve many problems, and help us accelerate the development of simulation systems. speed.

We were very rigorous when building this product. We built an automatic driving simulation system in accordance with the very standardized process of the automotive industry to verify the safety of automatic driving. We use game technology and data-driven excellence to build this system to help self-driving cars test and land.

02

Gaming Technology Makes Digital Twins

Create greater productivity

Later, we will expand the capabilities of digital twins to more industries, not only to solve the problem of autonomous driving simulation testing, but to make it more general.

Digital twins have been popular in the past few years. It consists of many modules, including rendering, big data, simulation, real-time interaction, etc. We will use the capabilities accumulated in simulation and game technology for a wider range of digital twin systems Construction, the core appeal is to help us better map the physical world to the twin world, and very low-cost trial and error, feedback the optimal solution or optimal plan to the physical world, and help us command or control the physical world .

What kind of game technology points did we use? One is the self-developed game engine. The game engine is the very core of the entire simulation twin. Using the self-developed game engine can achieve localization, and can help us customize the engine, instead of using a standardized commercial engine, commercial engine The heavy weight of the game will bring problems such as relatively high cost of calculation and low operating efficiency, and the self-developed game engine can help us perform simulation and rendering calculations more efficiently.

The second is automated modeling and editing tools. We use many such tools when we are doing digital twins, such as highly automated use of satellite images and map data to generate all roads, buildings, vegetation, and waters in any city within a day. Generated purely automatically.

There are also some multi-user collaborations. The glory of kings or console games that everyone plays essentially require multi-user collaboration for real-time interaction, and has higher requirements for data synchronization, requiring real-time data frame synchronization. Large-scale cloud rendering benefits from cloud games, which can help users not rely on the performance of their own terminals, but access the cloud to obtain a good rendering interactive experience. We do not tend to require users to purchase very expensive terminal equipment to run high-fidelity effects, so multi-user collaboration and cloud rendering are very core capabilities in digital twins.

What are we doing in the industry? For example, in the transportation industry, in the ground traffic simulation system, based on real-time traffic flow data, we use high-performance sensors, cameras + millimeter-wave radars to perceive real-time traffic flow reporting systems. It is not a purely virtual traffic flow scene, but a real-time map of smart cities. traffic.

Based on such traffic flow, we can inject a virtual event, such as a virtual traffic accident, and how it will affect the road, so as to help the traffic department to better guide. On the right is the application of drones, such as channel analysis, conflict prediction, resolution, etc. In the transportation industry, we have used the ability of game simulation to achieve the purpose of improving the efficiency of the transportation system.

Examples of our twin technology applications in different industries. For example, the government twin system and the factory twin system can improve factory efficiency and reduce personnel costs. The building twin application can connect elevators and all equipment for dynamic management, as well as intelligent control of air conditioners to reduce energy consumption…

We also build a carbon emission management platform for government personnel. Carbon emission is a relatively complicated issue. The industries with the largest carbon emission are industry, transportation, and construction. Therefore, to make a better carbon emission model, it is necessary to combine these three Carbon emission data and short-term predictions in various fields are very accurately twinned into the digital twin world, and future predictions are made based on carbon emission models to help our environmental managers calculate better solutions and better To achieve the purpose of carbon peak and carbon neutrality.

In summary, whether it is a game engine, automated modeling, or cloud rendering, they are all tools that provide users with mapping from the physical world to the twin world on the basis of a general-purpose digital twin.

This article is transferred from: https://www.geekpark.net/news/313640
This site is only for collection, and the copyright belongs to the original author.