Intelligent driving data solution 2.0: AI data processing efficiency is comprehensively improved, and cloud measurement data release technology is upgraded.

On August 24, the China International Autonomous Driving Expo was held in Shanghai. 10 conferences were held at the exhibition site, with a total of 120 keynote speeches, covering intelligent driving radar, visual cameras, maps and positioning, chassis domain control and parking and parking integration, information security and functional safety, AI data, simulation software, intelligent network Discussions on 10 topics including connectivity, smart cockpit, and low-speed scene solutions

The general manager of Cloud Test Data participated in the 2023 ICVS Intelligent Cockpit Technology Summit, which was also attended by FAW Group, Harman China, Xpeng Motors, Quanji Technology, Tongji University, China Society of Automotive Engineers, Beijing University of Technology and other industries Excellent representative. At the meeting, Cloud Test Data released a newly upgraded intelligent driving AI data solution 2.0, which provides the optimal AI data service solution for the development of intelligent driving and releases new momentum of high-efficiency and high-quality AI data

Under the dual influence of policies and large-scale artificial intelligence models, the process of intelligentization and commercialization of autonomous driving is accelerating. Since the scene data required for autonomous driving is long-tail data, it needs to cover as many edge situations as possible. Upgrading and iteration of algorithm models also requires continuous investment in new scene data, and the importance of data has become increasingly prominent. From automobile manufacturers to first-tier suppliers, various participants in the entire autonomous driving industry chain have begun to pay attention to the construction of a closed loop of autonomous driving data

Faced with the trend of gradual implementation of intelligent driving and more abundant application scenarios, data closed loop has become the current industry development trend. How to effectively circulate training data has become the key to improving the efficiency of related enterprises. It is reported that compared with the 1.0 version, the Cloud Measurement Data Intelligent Driving AI Data Solution 2.0 has been comprehensively upgraded in many aspects such as data closed-loop capabilities, automatic annotation capabilities, data management tool chains, and manual performance evaluation, with the integrated data base as the core. . This can comprehensively improve the efficiency of data annotation and circulation while ensuring the quality of data annotation

In order to realize a complete closed-loop model, we need large-scale, high-quality, multi-scenario data, high computing power, high efficiency, relatively low-cost algorithm models, tending towards automated data annotation and processing levels, high speed and low cost. transmission rate and storage mode, as well as security and compliance guarantees. With the input of new data, the closed-loop model can continuously cycle forward to realize the automated growth of autonomous driving. At present, autonomous driving solution providers such as Haomo Zhixing and Baidu Apollo have launched unique data closed-loop solutions. They spare no effort in data collection logic, data true value screening, training scenario simulation, algorithm model iteration, etc., and strive to achieve low Cost-effective, large-scale, high-quality, and efficient data closed loop

CITIC Securities research report pointed out that as advanced autonomous driving technology begins mass production in urban areas, the system's perception, decision-making, execution and communication architecture will usher in a new upgrade. At present, all participants use "BEV Transformer data closed loop" as the core architecture of the latest generation of autonomous driving mass production systems. The algorithm paradigm and technical routes are temporarily consistent. Data volume and data closed loop capabilities may become the next stage from 1 to N. The essential

The above is the detailed content of Intelligent driving data solution 2.0: AI data processing efficiency is comprehensively improved, and cloud measurement data release technology is upgraded.. For more information, please follow other related articles on the PHP Chinese website!