Research on five typical scenario automatic recognition methods based on lidar target data

Traditional testing methods no longer meet the requirements of ICV testing. Scenario based testing can perfectly solve this problem. The testing scenarios must come from real traffic data, and how to extract valuable scenarios from the massive actual collected data is a key issue. This paper proposes an automated scenario extraction method that accurately identifies five typical scenarios based on LiDAR target data. Based on the fragments of the right of way competition, determine whether the different key parameters of various scenarios have reached the threshold, in order to intercept and output valuable typical scenarios. Finally, the extraction results are verified by stratified sampling method, and the scenario recognition accuracy of the data segment is obtained by weighted calculation. The validation results indicate that the method designed in this paper has extremely high extraction accuracy and can effectively and correctly extract five typical scenarios.