Browsing by Author "Zhao, Yun"

Now showing 1 - 6 of 6
  • Thumbnail Image
    Item
    3D Object Detection Algorithm Based on the Reconstruction of Sparse Point Clouds in the Viewing Frustum
    (Hindawi Publishing Corporation, 2022-10-15) Xu, Xing; Wu, Xiang; Zhao, Yun; Lü, Xiaoshu; Aapaoja, Aki; Department of Civil Engineering; Structures – Structural Engineering, Mechanics and Computation; Zhejiang University of Science and Technology; Solita Oy
    In response to the problem that the detection precision of the current 3D object detection algorithm is low when the object is severely occluded, this study proposes an object detection algorithm based on the reconstruction of sparse point clouds in the viewing frustum. The algorithm obtains more local feature information of the sparse point clouds in the viewing frustum through dimensional expansion, performs the fusion of local and global feature information of the point cloud data to obtain point cloud data with more complete semantic information, and then applies the obtained data to the 3D object detection task. The experimental results show that the precision of object detection in both 3D view and BEV (Bird's Eye View) can be improved effectively through the algorithm, especially object detection of moderate and hard levels when the object is severely occluded. In the 3D view, the average precision of the 3D detection of cars, pedestrians, and cyclists at a moderate level can be increased by 7.1p.p., 16.39p.p., and 5.42p.p., respectively; in BEV, the average precision of the 3D detection of car, pedestrians, and cyclists at hard level can be increased by 6.51p.p., 16.57p.p., and 7.18p.p., respectively, thus indicating the effectiveness of the algorithm.
  • No Thumbnail Available
    Item
    Giant enhancement of optical nonlinearity in two-dimensional materials by multiphoton-excitation resonance energy transfer from quantum dots
    (Nature Publishing Group, 2021-07) Hong, Hao; Wu, Chunchun; Zhao, Zixun; Zuo, Yonggang; Wang, Jinhuan; Liu, Can; Zhang, Jin; Wang, Fangfang; Feng, Jiangang; Shen, Huaibin; Yin, Jianbo; Wu, Yuchen; Zhao, Yun; Liu, Kehai; Gao, Peng; Meng, Sheng; Wu, Shiwei; Sun, Zhipei; Liu, Kaihui; Xiong, Jie; Department of Electronics and Nanoengineering; Centre of Excellence in Quantum Technology, QTF; Zhipei Sun Group; University of Electronic Science and Technology of China; Peking University; CAS - Institute of Physics; Beijing Institute of Technology; Henan University; CAS - Technical Institute of Physics and Chemistry; Beijing Graphene Institute; Fudan University
    Colloidal quantum dots are promising photoactive materials that enable plentiful photonic and optoelectronic applications ranging from lasers, displays and photodetectors to solar cells1–9. However, these applications mainly utilize the linear optical properties of quantum dots, and their great potential in the broad nonlinear optical regime is still waiting for full exploration10–12. Here, we demonstrate that a simple coating of a sub-200-nm-thick quantum dot film on two-dimensional materials can significantly enhance their nonlinear optical responses (second, third and fourth harmonic generation) by more than three orders of magnitude. Systematic experimental results indicate that this enhancement is driven by a non-trivial mechanism of multiphoton-excitation resonance energy transfer, where the quantum dots directly deliver their strongly absorbed multiphoton energy to the adjacent two-dimensional materials by a remote dipole–dipole coupling. Our findings could expand the applications of quantum dots in many exciting areas beyond linear optics, such as nonlinear optical signal processing, multiphoton imaging and ultracompact nonlinear optical elements.
  • No Thumbnail Available
    Item
    Optical fibres with embedded two-dimensional materials for ultrahigh nonlinearity
    (Nature Publishing Group, 2020-12) Zuo, Yonggang; Yu, Wentao; Liu, Can; Cheng, Xu; Qiao, Ruixi; Liang, Jing; Zhou, Xu; Wang, Jinhuan; Wu, Muhong; Zhao, Yun; Gao, Peng; Wu, Shiwei; Sun, Zhipei; Liu, Kaihui; Bai, Xuedong; Liu, Zhongfan; Department of Electronics and Nanoengineering; Centre of Excellence in Quantum Technology, QTF; Zhipei Sun Group; Chinese Academy of Sciences; Peking University; Beijing Institute of Technology; Fudan University
    Nonlinear optical fibres have been employed for a vast number of applications, including optical frequency conversion, ultrafast laser and optical communication1–4. In current manufacturing technologies, nonlinearity is realized by the injection of nonlinear materials into fibres5–7 or the fabrication of microstructured fibres8–10. Both strategies, however, suffer from either low optical nonlinearity or poor design flexibility. Here, we report the direct growth of MoS2, a highly nonlinear two-dimensional material11, onto the internal walls of a SiO2 optical fibre. This growth is realized via a two-step chemical vapour deposition method, where a solid precursor is pre-deposited to guarantee a homogeneous feedstock before achieving uniform two-dimensional material growth along the entire fibre walls. By using the as-fabricated 25-cm-long fibre, both second- and third-harmonic generation could be enhanced by ~300 times compared with monolayer MoS2/silica. Propagation losses remain at ~0.1 dB cm–1 for a wide frequency range. In addition, we demonstrate an all-fibre mode-locked laser (~6 mW output, ~500 fs pulse width and ~41 MHz repetition rate) by integrating the two-dimensional-material-embedded optical fibre as a saturable absorber. Initial tests show that our fabrication strategy is amenable to other transition metal dichalcogenides, making these embedded fibres versatile for several all-fibre nonlinear optics and optoelectronics applications.
  • No Thumbnail Available
    Item
    Research on global path planning algorithm for mobile robots based on improved A*
    (Elsevier Ltd, 2024-06-01) Xu, Xing; Zeng, Jiazhu; Zhao, Yun; Lü, Xiaoshu; Department of Civil Engineering; Performance in Building Design and Construction; Zhejiang University of Science and Technology
    In order to shorten searching time and reduce the quantity of redundant nodes in path planning, an improved A* algorithm was proposed. In the novel algorithm, compared with the A* algorithm, the octet neighborhood was replaced a rectangular boundary without obstacles. The map was explored in two directions from the starting point and the target point respectively. The exploration generated fewer nodes. Due to the enlargement of the search neighborhood and the rectilinear passage without obstacles in the rectangular region, the new algorithm used Euclidean distance as distance estimate. And a new operator was designed to seek the best search node to reduce the time complexity, so as to improve the efficiency of the algorithm. For improving the safety of mobile robot, the novel algorithm adopted adaptive cost function to improve the ability of road safety discrimination. A Slide-Rail corner adjustment method was designed to reduce unnecessary corners and improve the smoothness of the path. The simulation results showed that, compared with the traditional A* algorithm and various improved A* algorithms, the proposed algorithm can shorten the path length and searching time, reduce the number of turns, the total turning angles and search nodes. Compared to A*, the searching time was reduced by 64.76%, the total turning angles were reduced by 56.34%, and the search nodes were reduced by 82.76% averagely in test maps of this paper. Moreover, the average interval width in improved A* was 3.07 and was more than 1.5 times the average interval width in Rectangle Expansion A*.
  • No Thumbnail Available
    Item
    Short-term traffic flow prediction based on whale optimization algorithm optimized BiLSTM_Attention
    (JOHN WILEY & SONS, 2022-05-01) Xu, Xing; Liu, Chengxing; Zhao, Yun; Lv, Xiaoshu; Zhejiang University of Science and Technology; Structures – Structural Engineering, Mechanics and Computation; Department of Civil Engineering
    With the growths in population and vehicles, traffic flow becomes more complex and uncertain disruptions occur more often. Accurate prediction of urban traffic flow is important for intelligent decision-making and warning, however, remains a challenge. Many researchers have applied neural network methods, such as convolutional neural networks and recurrent neural networks, for traffic flow prediction modeling, but training the conventional network that can obtain the best network parameters and structure is difficult, different hyperparameters lead to different network structures. Therefore, this article proposes a traffic flow prediction model based on the whale optimization algorithm (WOA) optimized BiLSTM_Attention structure to solve this problem. The traffic flow is predicted first using the BiLSTM_Attention network which is then optimized by using the WOA to obtain its four best parameters, including the learning rate, the training times, and the numbers of the nodes of two hidden layers. Finally, the four best parameters are used to build a WOA_BiLSTM_Attention model. The proposed model is compared with both conventional neural network model and neural network model optimized by the WOA. Based on the evaluation metrics of MAPE, RMSE, MAE, and R2, the WOA_BiLSTM_Attention model proposed in this article presents the best performance.
  • No Thumbnail Available
    Item
    Urban short-term traffic speed prediction with complicated information fusion on accidents
    (Elsevier Ltd, 2023-08-15) Xu, Xing; Hu, Xianqi; Zhao, Yun; Lü, Xiaoshu; Aapaoja, Aki; Department of Civil Engineering; Performance in Building Design and Construction; Zhejiang University of Science and Technology; Solita Plc.
    Optimizing the traffic flow prediction system is crucial in developing intelligent transportation since it increases the road network's capacity. The system's overall prediction accuracy will be increased by taking into account the relationship between the temporal and spatial properties of the road network and different external elements affecting the traffic situation. The traffic state, which is still a largely unexplored area, is impacted by the complicated interaction between accident information and the spatiotemporal properties of the route. This paper proposes an Accident Information Graph Fusion Attention Convolutional Network(AI-GFACN). Firstly, a highly correlated global road network is created using a global spatial feature point-edge swapping method, a D–D algorithm fusing Dijkstra, and Depth-First Search, which resolves the issue where the spatial features of accident sections are challenging to capture the diffusion effects caused by spatial features of nearby and further sections. Following the data's incorporation, it is suggested to combine the Spatio-temporal features of accident information and embed them in the road network. In addition, an attention mechanism is introduced, effectively addressing the difficulty in capturing the Spatio-temporal features of accident information within the road network. By integrating and categorizing the regionally distributed and temporally sustained congestion effects of various categories of accidents concerning previous research on accident information, this paper enhances the semantic expressiveness of accident information within the road network. Ablation experiments confirm the effectiveness and robustness of the proposed method, and it is applied to the dataset of Hangzhou West Lake District (including accident information), which increases short-term traffic speed prediction accuracy by 0.2% overall.