Baidu releases closed-loop compliance plan for autonomous driving data to drive the safe development of the autonomous driving industry

On November 2, Baidu Apollo’s “Autonomous Driving Cloud 2.0 Pre-release Communication Conference” was held in Yizhuang, Beijing. It was disclosed at the meeting that Baidu will provide auto companies with a full-stack data closed-loop compliance solution for autonomous driving research and development, to meet the safety compliance requirements for smart car and autonomous driving data collection, and to assist auto companies in the development, operation, and commercialization of full-scenario closed-loop solutions. The automobile industry injects data-driven power, empowers car companies to create innovative smart driving applications, and promotes the safe and rapid implementation of autonomous driving scenarios. In addition, it is understood that Baidu’s autonomous driving cloud 2.0 panorama will also be officially released in December.

740

(Baidu Autonomous Driving Cloud 2.0 pre-release communication meeting site)

In the past few years, my country’s smart car industry has developed vigorously, with the increasing ability of automobile data processing, the increasing scale of automobile data, and the increasingly prominent problems of automobile data security and hidden risks. At the end of August this year, the Ministry of Natural Resources issued the “Notice on Promoting the Development of Intelligent and Connected Vehicles and Maintaining the Security of Surveying and Mapping Geographic Information”, which put forward clear requirements for the surveying and mapping subjects and the development of surveying and mapping activities of intelligent and connected vehicles. Under the guidance of the policy, the behavior of car companies in the process of R&D and mass production of intelligent networked vehicles and autonomous vehicles is gradually becoming compliant. However, at the same time, the three major challenges of data compliance for autonomous driving still exist, data rights confirmation, data protection, and data circulation, and a large number of car companies are in urgent need of relevant technologies, experience and qualifications, and compliance services have become the focus of industry development.

At the communication meeting, Liu Jianhao, head of information security at Baidu apollo, introduced the technical solution of Baidu Autopilot Cloud 2.0 in terms of closed-loop autopilot data. He said that Baidu Autopilot Cloud and Baidu Zhitu, adhering to the safety and compliance idea of ​​”no original data in the car, no mapping data in the cloud, unrelated surveying and mapping results, and full management and control of qualification map business”, help car companies to meet the closed loop of autonomous driving data. Compliance.

In order to implement the compliance principle, Baidu has proposed the security compliance construction concept of “one center, triple protection” to ensure the security of geographic information. One center is to establish a data security supervision center through the map dealer to monitor the data usage process, data content, and safety of data carriers for closed-loop compliance of autonomous driving data. Fully meet the safety compliance requirements of regulatory agencies for smart cars and autonomous driving.

The triple protection refers to tool chain compliance, data protection, and cloud platform protection.

Tool chain compliance: ensure the operation of autonomous driving data in the training process is compliant by means of point cloud frame extraction, image desensitization, controllable annotation, vehicle supervision, etc., without touching or taking away the surveying and mapping data.

Data protection: Carry out end-to-end encryption transformation for the spatiotemporal data of intelligent networked vehicles to ensure the integrity, validity, confidentiality, and non-repudiation of the data, and ensure that the data is managed and controlled by the map provider through the cryptographic service system.

Cloud platform protection: Based on measures such as network boundary isolation, container protection, data auditing, and secure storage of cloud infrastructure, the security of intelligent networked vehicle spatiotemporal data in the cloud platform environment is ensured. The surveying and mapping results will not be sent out, causing risks such as data abuse and leakage, to resist hackers’ attacks on the data platform and prevent data from being stolen by hackers.

“In the future, we will continue to develop in three aspects, including data compliance hosting services, compliance proprietary cloud services, and autonomous driving R&D tool chain services, to provide car companies with higher-quality autonomous driving technical capabilities,” said Liu Jianhao.

It is reported that Baidu’s self-driving technology research and development capabilities have been accumulated for more than 10 years. The cumulative self-driving mileage exceeds 34 million kilometers, which can effectively support thousands of petabytes of mass storage, 10,000 elastic computing clusters, and the collection, transmission and storage of surveying and mapping data. All of them have been practiced a lot. Therefore, Baidu’s autonomous driving cloud 2.0 compliance solution not only has the advantages of full stack, but also has the characteristics of efficient iteration and massive scene data. It can build a closed loop of vehicle cloud data based on user feedback to quickly improve the effect.

At the same time, the large number of qualifications obtained by Baidu Apollo in the past ten years are also an important basis for its ability to build closed-loop compliance of full-stack autonomous driving data. Baidu is not only one of the 19 companies in the country with Class-A surveying and mapping qualifications for navigation electronic map production, but also has the recognition of network security level protection and trusted cloud. In terms of vehicle end landing, Baidu has professional automotive network security engineering capability certification, functional safety certification, privacy protection certification and other comprehensive safety qualifications, enabling Baidu to combine various qualifications on the basis of surveying and mapping entities to issue compliance collection plans to ensure vehicle security. Enterprise mass production projects can achieve end-to-end compliance.

In the digital age, autonomous driving has come to the forefront, giving birth to new kinetic energy in technological evolution and travel reform, bringing tangible changes to people’s travel life. In the future, we will continue to explore the safety and compliance development of autonomous driving, in-depth service for car companies to upgrade their safety compliance needs, and continuously strengthen the construction of autonomous driving infrastructure to promote the expansion of the development space of the autonomous driving industry.

This article is reproduced from: https://www.leiphone.com/category/industrynews/DYC0o6V1gKjL5Lrg.html
This site is for inclusion only, and the copyright belongs to the original author.