我所李润梅老师课题组完成的论文《A Vehicle Behavior Prediction Method Based on CNN-Transformer Model》于2024-01在《Proceedings of 2024 International Conference on Computer Vision and Deep Learning(CVDL)》正式发表。论文摘要如下：
Abstract: Accurately predicting of the driving behavior of adjacent vehicles is one of the things that need to be done before the control decision of unmanned driving, and is one of the key factors of driving safety. We propose a CNN-Transformer combined model based on a data-driven approach to accurately predict driving behavior of moving vehicles. The first module is a Convolutional Neural Network, which is used to extract spatial features to capture behavioral characteristics of the vehicle. In addition, another module in the model is the Transformer, which is used to process time series data to capture the interaction characteristics between vehicles. Through training and optimization of model parameters, a deep model can accurately predict driving behavior, while reducing overfitting and information loss problems. Finally, the simulation results show that the CNN-Transformer combination model has excellent performance in vehicle behavior prediction.
引用信息：Xiaoting Shu, Runmei Li. A Vehicle Behavior Prediction Method Based on CNN-Transformer Model. Proceedings of 2024 International Conference on Computer Vision and Deep Learning(CVDL), 2024-01. DOI: 10.1145/3653781.3653797