Analysis of key components and technical trends of autonomous driving simulation platform

The Defense Acquisition University of the United States defines digital twins as follows: Digital twins make full use of physical models, sensor updates, operation history and other data, integrate multi-disciplinary, multi-physical quantity, multi-scale, and multi-probability simulation processes, complete mapping in virtual space, and thus reflect the entire life cycle of the corresponding physical equipment.

Digital twins, also known as "digital twins", digitally copy a physical object and simulate the behavior of the object in the real environment; that is, create an effective "replica". Just like the real and fake "Monkey King" in Journey to the West, the two objects have almost the same attributes, not only the same appearance, but even the same magic skills.

For autonomous driving simulation testing, the typical application of digital twin technology is the autonomous driving digital twin test VRIL (Virtual Reality in the Loop); that is, using digital twin technology to create a virtual environment model that is consistent with the real world. While the real vehicle is tested in a real test site or road, it will be synchronously mapped to the virtual test environment, thus forming a closed-loop test of the entire vehicle that combines virtual and real.

Autonomous driving test system based on scenario library (picture from the Internet)

The virtual environment has the same roads, traffic participants, test vehicles and sensor models as in the real world. The various motion state information of the real-world test vehicles in the real environment and the environmental information collected by the sensors will be updated synchronously in the virtual environment.

At the same time, the target information detected by the sensor model in the virtual environment can also be fed back to the test vehicle in the real world for information fusion and auxiliary decision-making. Therefore, the interaction of information and status between the real world and the virtual environment is realized, and the closed-loop real-time simulation test of the digital twin system is realized.

3.3 Mixed traffic flow simulation

It is not a one-off thing for self-driving vehicles to replace traditional cars. They will inevitably coexist for quite some time, so the mixed traffic of traditional cars and self-driving cars will be a common traffic scenario. Xu Changming, director of the National Information Center, once pointed out at a conference that the biggest difficulty in China's development of L4 and L5 self-driving cars is "mixed traffic", that is, the mixed traffic of self-driving cars and non-self-driving cars.

Example of urban mixed traffic flow micro-simulation system (picture from the Internet)

Relevant experts have suggested that the key to solving this problem is to realize the transformation of autonomous driving vehicles from individual intelligence to group intelligence, that is, the integrated vehicle-road-cloud networked collaborative intelligence led by my country - realizing beyond-line-of-sight perception and advance prediction through networking, thereby reducing the pressure of individual vehicles at the environmental perception level; centralized decision-making through the cloud brain, reducing the pressure of individual vehicles at the decision-making level, and improving the efficiency and reliability of decision-making.

Whether the safety issues of autonomous vehicles in mixed traffic conditions can be solved will be a key factor in determining whether autonomous vehicles can be commercialized. Test scenarios based on mixed traffic conditions (especially V2X scenario testing under mixed traffic conditions) will be an important research direction for autonomous driving simulation testing in the future, and even simulation testing of mixed traffic models with different penetration rates of autonomous vehicles is required.

References:

1. Automobile Intelligent Driving Simulation Technology (Book) | Machinery Industry Press

2. Blue Book on Simulation Testing of Autonomous Driving Vehicles (Book) | Publishing House of Electronics Industry

3. China Autonomous Driving Simulation Technology Research Report 2019 (Report)

4. 2020 China Autonomous Driving Simulation Blue Book (Report)

5. Cross-border thinking in traffic engineering: What are the available platforms for traffic flow simulation in unmanned driving simulation?

https://mp.weixin.qq.com/s/Y0qN-EuQqWBwsyxbNpBTQA

6. Overview of Traffic Flow Simulation

https://mp.weixin.qq.com/s/Phikehg4UuLQp32w6ISLhw

7. Digital Twins and Autonomous Driving Testing

https://www.sohu.com/a/417352736_468661

8. Overview of research on the application of digital twin technology in the field of autonomous driving testing

https://www.auto-tesTIng.net/news/show-110318.html

9. Sensor Model in Autonomous Driving Simulation System

https://zhuanlan.zhihu.com/p/66963787

10. Simulation testing is the only way to autonomous driving | Houshi Automobile

https://www.sohu.com/a/241962244_465591

11. Breakthrough in the second half of autonomous driving: Virtual simulation core technology must be autonomous and controllable

https://finance.sina.com.cn/roll/2020-10-14/doc-iiznezxr5919819.shtml

12. Zhang Fan: Simulation testing is the most basic tool to ensure automobile safety | China Auto News

https://mp.weixin.qq.com/s/JwsDYF1jRGmnGz9nuh7xSw

13. Li Manman: The mission of simulation technology in the era of autonomous driving

https://www.fangzhenxiu.com/post/1853946

14. Corner cases in autonomous driving

https://mp.weixin.qq.com/s/CGzbYTIrm6iajnbX-FADJg

15. Three Apache frameworks commonly used for processing large data streams: Storm, Spark, and Samza

https://www.cnblogs.com/myinspire/p/7273125.html

16. Introduction to Virtual Simulation Testing: Introduction to Hardware-in-the-Loop (HIL) Testing

https://mp.weixin.qq.com/s/CeSzo4SJitcQ-YcYavbEvA

17. Driver-in-the-Loop (DIL): Virtual Reconstruction and Twin Verification

https://mp.weixin.qq.com/s/_wYjpmV8yGwiHc77qBL_lA

18. Research on scenario-based virtual simulation testing of autonomous vehicles

https://zhuanlan.zhihu.com/p/85613284

19. Simulation test! Uncovering the new trend of autonomous driving testing under Vision Zero

https://www.auto-testing.net/news/show-111494.html