关键词: 智能网联汽车；数据闭环；数据采集；感知测试

Abstract: To address the functional anomalies and deficiencies in autonomous driving systems caused by the performance limitations of onboard sensors such as cameras and radar, this study aims to establish a universal sensor testing and evaluation framework. This framework is intended to provide a basis for OEMs' sensor selection and performance validation. Leveraging practical experience in data-driven closed-loop engineering, a comprehensive theoretical and methodological system is constructed. This system covers the entire process from in-vehicle configuration and data acquisition to data processing and perceptual performance testing, thereby elevating sensor testing from isolated functional verification to systematic and quantifiable performance evaluation. The research resulted in a complete and actionable methodology for sensor testing and evaluation. This framework enables the systematic assessment of key sensor performance indicators, accurately revealing their performance boundaries and limitations. The proposed methodological system provides a universal testing standard for the industry. It can not only effectively guide OEMs in sensor selection decisions and enhance the reliability of autonomous driving systems but also holds significant reference value for advancing the development of environmental perception technology across the sector.

Key words: intelligent and connected vehicles; data closed loop; data collection; perception testing