Parameter-Efficient Vehicle Trajectory Prediction Based on Attention-Enhanced Liquid Structural Neural Model DOI Creative Commons
Ruochen Wang, Yue Chen, Renkai Ding

и другие.

World Electric Vehicle Journal, Год журнала: 2024, Номер 16(1), С. 19 - 19

Опубликована: Дек. 31, 2024

Due to advances in sensor techniques and deep learning, autonomous vehicular technologies have become more reliable practical. Trajectory prediction is a critical task anticipate the future positions of surrounding vehicles. However, existing algorithms, such as LSTM-based attention-based models, face challenges high computational complexity, large parameter sizes, limited ability efficiently capture both temporal dependencies spatial interactions dynamic traffic scenarios. In this paper, we propose parameter-efficient trajectory model that integrates Liquid Time-Constant (LTC) networks with attention mechanisms, termed Attn-LTC model. The key contributions our work are threefold. First, introduce attention-enhanced LTC encoder effectively captures long-term behaviors from historical data. Second, incorporate decoder, which emphasizes influence neighboring vehicles interactions, thereby improving accuracy. Third, demonstrate efficiency model, achieves predictive accuracy significantly fewer parameters compared Transformer-based counterparts. Extensive experiments conducted on NGSIM dataset advantages proposed Notably, it reduces complexity size while maintaining superior accuracy, making well suited for deployment resource-constrained systems. results highlight effectiveness balancing precision efficiency, paving way its application real-time driving

Язык: Английский

A review of multilayer networks-based interregional transportation networks analysis DOI
Jiaqi Li,

Zhenfu Li,

Xinli Qi

и другие.

Chaos Solitons & Fractals, Год журнала: 2025, Номер 192, С. 115993 - 115993

Опубликована: Янв. 16, 2025

Язык: Английский

Процитировано

0

Motion-Inspired Spatial–Temporal Transformer for accurate vessel trajectory prediction DOI
Huimin Qiang, Zhiyuan Guo, Zhong Chu

и другие.

Engineering Applications of Artificial Intelligence, Год журнала: 2025, Номер 148, С. 110391 - 110391

Опубликована: Март 6, 2025

Язык: Английский

Процитировано

0

Skip or not: Hybrid machine learning for decision support in strategic port-skipping behavior to enhance liner shipping reliability DOI

Xingcan Fan,

Jing Lyu, Lingye Zhang

и другие.

Ocean Engineering, Год журнала: 2025, Номер 324, С. 120730 - 120730

Опубликована: Фев. 22, 2025

Язык: Английский

Процитировано

0

Attention-enhanced and integrated deep learning approach for fishing vessel classification based on multiple features DOI Creative Commons
Xin Cheng, Jintao Wang, Xinjun Chen

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Март 13, 2025

Effective fisheries management is the key to achieve sustainable globally, while accurate monitoring of fishing vessels essential improve effectiveness measures. Self-reported information on vessel types often limited and may not cover all operating vessels, causing incomplete in management. Therefore, a novel way objectively identify large quantity needed. In this study, we presented an innovative integrated deep learning model by using automatic identification system (AIS) data classify five including gillnetter, hook liner, trawler, fish carrier, stow net vessel, further improving performance classification. First, preprocessed removing erroneous information, dividing trajectories day obtain complete reliable dataset. Then, multidimensional feature vector was constructed combining geometric, static dynamic characteristics explain behavioral differences various more effectively. Finally, fed into ensemble two-dimensional bidirectional long short-term memory network convolutional neural with attention mechanism for training, prediction results were obtained through fully connected layer. The accuracy 91.90%, which higher than other single classifiers. experimental demonstrated that method remarkable could be adopted precision classification based AIS data.

Язык: Английский

Процитировано

0

Research on ship dynamic feature extraction and prediction method based on visual data DOI
Stephen Lin, Yong Li, Yu Hu

и другие.

Ocean Engineering, Год журнала: 2025, Номер 327, С. 120938 - 120938

Опубликована: Март 17, 2025

Язык: Английский

Процитировано

0

Deep multi-view information-powered vessel traffic flow prediction for intelligent transportation management DOI Creative Commons
Huanhuan Li, Yu Zhang,

Yan Li

и другие.

Transportation Research Part E Logistics and Transportation Review, Год журнала: 2025, Номер 197, С. 104072 - 104072

Опубликована: Март 21, 2025

Язык: Английский

Процитировано

0

A framework for ship semantic behavior representation and indexing DOI

Shunqiang Xu,

Liang Huang, Yamin Huang

и другие.

Ocean Engineering, Год журнала: 2025, Номер 329, С. 121023 - 121023

Опубликована: Апрель 7, 2025

Язык: Английский

Процитировано

0

BESO-PPF: A PPF-optimized ship heading controller based on backstepping control and the ESO DOI
Changqing Wang, Xiaori Gao, Lidong Wang

и другие.

Ocean Engineering, Год журнала: 2024, Номер 316, С. 119925 - 119925

Опубликована: Дек. 3, 2024

Язык: Английский

Процитировано

0

Parameter-Efficient Vehicle Trajectory Prediction Based on Attention-Enhanced Liquid Structural Neural Model DOI Creative Commons
Ruochen Wang, Yue Chen, Renkai Ding

и другие.

World Electric Vehicle Journal, Год журнала: 2024, Номер 16(1), С. 19 - 19

Опубликована: Дек. 31, 2024

Due to advances in sensor techniques and deep learning, autonomous vehicular technologies have become more reliable practical. Trajectory prediction is a critical task anticipate the future positions of surrounding vehicles. However, existing algorithms, such as LSTM-based attention-based models, face challenges high computational complexity, large parameter sizes, limited ability efficiently capture both temporal dependencies spatial interactions dynamic traffic scenarios. In this paper, we propose parameter-efficient trajectory model that integrates Liquid Time-Constant (LTC) networks with attention mechanisms, termed Attn-LTC model. The key contributions our work are threefold. First, introduce attention-enhanced LTC encoder effectively captures long-term behaviors from historical data. Second, incorporate decoder, which emphasizes influence neighboring vehicles interactions, thereby improving accuracy. Third, demonstrate efficiency model, achieves predictive accuracy significantly fewer parameters compared Transformer-based counterparts. Extensive experiments conducted on NGSIM dataset advantages proposed Notably, it reduces complexity size while maintaining superior accuracy, making well suited for deployment resource-constrained systems. results highlight effectiveness balancing precision efficiency, paving way its application real-time driving

Язык: Английский

Процитировано

0