Generalization Generation of Dangerous Lane-Changing Scenarios for Automated Vehicle Testing

To address the issue of dangerous lane-changing scenario construction in automated vehicle virtual testing, this paper proposed a data-model-driven method for generally producing dangerous lane-changing scenarios. Based on emergency lane-changing data in the NGSIM US101 dataset, an emergency lane-changing trajectories producing method (BN-AN-SeqGAN) is proposed based on a sequence generative adversarial network. The safety distance-based constraint model for two-vehicle lane-changing states is built, and the general approach of producing dangerous lane-changing test scenarios is designed. The library of dangerous lane-changing test scenarios is finally achieved. According to the experimental findings, the root mean square error of the lane-changing completion time distribution for produced 50,000 emergency trajectories is 0.63. Among the 50,000 generated dangerous lane-changing scenarios, time to collision between the tested automated vehicle and the lane-changing background vehicle is less than 1 s in 99.54% of the scenarios. The results show that the proposed method can effectively produce dangerous lane-changing scenarios for automated vehicle testing.