Research on Test and Evaluation Method of L3 Intelligent Vehicle Based on Chinese Characteristics Scene

Abstract

With the development of autonomous driving technology, the L3 intelligent and connected vehicle gradually takes the vector production stage, and test verification is an essential part to measure its functionality. There is no relevant guidance method at present. Based on the natural driving scene database of China Automotive Technology and Research Center, this paper uses LGBM decision tree model to extract the typical features of the scene, and outputs the final high complexity scene according to the weight of each tree. At the same time, combined with CIDAS accident scene and parameter recombination scenario, this paper proposes a kind of scene design method and evaluation method for L3 intelligent vehicle function test. Finally, the verification is carried out on the intelligent and connected vehicle VIL test platform. The results show that the method has good consistency with the actual vehicle test results.