大连理工大学学报2026，Vol.66Issue(2)：147-155,9.DOI:10.7511/dllgxb202602006
考虑驾驶风格的驾驶意图识别模型
Driving intention recognition model considering driving styles
摘要
Abstract
Accurate recognition of vehicle driving intention plays an important role in ensuring the safety of vehicle operation and enhancing the performance of autonomous vehicles.To address the issue that existing driving intention recognition models fail to fully consider driver diversity,a driving intention recognition model that incorporates the representation of driving styles for both the ego vehicle and surrounding vehicles is proposed.First,a driving behavior matrix(DBM)is constructed based on vehicle trajectory information,with implicit driving style features extracted through a convolutional neural network(CNN),and corresponding weight coefficients are used to measure the influence of surrounding vehicles' driving styles;to reduce redundant information in trajectory data,dimensionality reduction of trajectory information is performed based on interaction factors,and a bi-directional long short-term memory(Bi-LSTM)network is employed to capture the temporal relationships of interaction factors,thereby obtaining inter-vehicle interaction features;finally,inter-vehicle interaction features and driving style features are combined,and driving intention recognition is carried out through a fully connected network.Validation results on the NGSIM(next generation simulation)dataset show that the proposed model achieves a driving intention recognition accuracy of 98.11％,demonstrating better recognition performance compared to existing driving intention recognition models.关键词
智能汽车/驾驶意图识别/驾驶风格/卷积神经网络Key words
intelligent vehicle/driving intention recognition/driving style/convolutional neural network分类
交通工程引用本文复制引用
李俊涛,张明恒,王潇雨,伊兴海..考虑驾驶风格的驾驶意图识别模型[J].大连理工大学学报,2026,66(2):147-155,9.基金项目
国家自然科学基金资助项目(52272413). （52272413）
