Research on cut-in scenario identification algorithm based on risk assessment

Abstract

The boundaries of typical scenes need to be extracted from the original scene data occurring in real roads, but the manual scene labeling method is inefficient and costly. Therefore, it is important to study the acquisition of autonomous driving scenes in real roads for the testing of autonomous vehicles, and the cut-in scenes are typical hazard scenes. In this paper, a scenario acquisition system with multi-source heterogeneous sensors is modified based on a vehicle equipped with L2- level autonomous driving system for acquiring natural road driving scenario data. Subsequently, a cut-in scenario automatic recognition algorithm based on risk assessment is proposed, and a recognition algorithm evaluation test platform is constructed, and a large number of test case generation procedures are designed to simulate and test the algorithm. And a large amount of natural driving data is used to verify the proposed recognition algorithm. The results show that the algorithm proposed in this paper can effectively identify the cut-in scenes occurring in front of the collection vehicle and has high accuracy.