检索结果-内蒙古大学图书馆

3rd International Conference on Robotics and Control engineering, RobCE 2023

作者： Chen, Liangji Wang, Zhiling Wang, Hanqi Zhou, Pengfei Hefei Institutes of Physical Science Chinese Academy of Sciences Anhui Hefei China University of Science and Technology of China Anhui Hefei China Anhui Engineering Laboratory for Intelligent Driving Technology and Application Anhui Hefei China

ISBN: (纸本)9781450398107

In the field of robotics and autonomous driving, achieving accurate 3D object detection is crucial for the perception of complex traffic environments. Most current research uses deep learning methods to extract object features from point clouds or images. However, these approaches often do not fully utilize the mutual positional information between objects, resulting in low detection accuracy in dense scenes. To address this issue, this paper proposes a frustum point cloud 3D target detection algorithm based on the fusion of camera and LiDAR data. We establish global connectivity of objects based on the degree of overlap between camera detection frames. Then, we design a neighborhood graph search algorithm based on constraint satisfaction to match the camera target detection results with LiDAR clustering results. Finally, the category and distance information of obstacles are displayed in bird's eye view (BEV). Evaluated on the KITTI benchmarks, our method achieves an average precision (AP) of 88.38% on the easy level, 87.64% on the moderate level, and 80.10% on the hard level in BEV detection. Compared to F-ConvNet, our method shows improvements of 5.49% on the moderate level and 7.33% on the hard level, significantly enhancing recognition accuracy in dense scenes. © 2023 Owner/Author.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Drivable Area Segmentation in Unstructured Roads for Autonomous Vehicles based on Multi-sensor Fusion 11

Drivable Area Segmentation in Unstructured Roads for Autonom...

引用

11th International Conference on Computing and Pattern Recognition, ICCPR 2022

作者： Wang, Hanqi Liang, Huawei Chen, Liangji Gong, Diancheng Zhou, Pengfei Kong, Bin Hefei Institutes of Physical Science Chinese Academy of Sciences Anhui Hefei China University of Science and Technology of China Anhui Hefei China Anhui Engineering Laboratory for Intelligent Driving Technology and Application Anhui Hefei China Chinese Academy of Sciences Anhui Hefei China

ISBN: (纸本)9781450397056

Drivable area segmentation is vital for autonomous vehicle driving safety, especially on unstructured roads. Mainstream drivable area algorithms are suited for structured environments, such as urban roads. However, these algorithms perform poorly in unstructured environments. This paper proposes a drivable area segmentation algorithm based on multi-sensor late-fusion for unstructured environments. The algorithm uses the visual segmentation results to correct the light detection and ranging (LiDAR) segmentation results, which can effectively solve those environments with unapparent boundary height differences. Desert experiments show that our algorithm achieves 96.02 on Intersection over Union (IoU), which is 36.75 and 38.31 higher than the LiDAR-based and the Vision-based algorithm, respectively. © 2022 ACM.

关键词： Autonomous vehicles

来源：评论

学校读者我要写书评

暂无评论

Simulation of a novel curved photovoltaic(PV)window improving the annual daylighting and building energy performance simultaneously

引用

Building Simulation 2024年第11期17卷 1951-1969页

作者： Fangfang Gong Yuan Gao Xinyi Tian Jun Wang Jie Ji Feng Shi Xiaoqiang Hong Fujian Province University Key Laboratory of Intelligent and Low-carbon Building Technology School of Architecture and Civil EngineeringXiamen UniversityXiamen361005FujianChina Fujian Key Laboratory of Digital Simulations for Coastal Civil Engineering Xiamen UniversityXiamen361005FujianChina Xiamen Key Laboratory of Integrated Application of Intelligent Technology for Architectural Heritage Protection Xiamen361005China Department of Thermal Science and Energy Engineering University of Science and Technology of ChinaHefei230026AnhuiChina School of Environment and Energy Engineering Anhui Jianzhu UniversityHefei230026AnhuiChina

Building integrated photovoltaic(BIPV)windows impact building performance by balancing daylighting availability,visual comfort,solar power generation,and building energy *** this balance is crucial for improving overall building energy efficiency and indoor environment *** study introduces a novel curved photovoltaic window design aimed at increasing daylight transmittance while maintaining the same photovoltaic area as a flat PV *** annual daylighting availability,visual comfort and building energy performance of three types of flat/curved PV windows(180°,120°,0°-flat)in a reference office room was comparatively studied across five different climate zones in China(Xiamen,Harbin,Nanjing,Kunming,and Beijing).The PV model was validated by the experimental *** results showed that the room with curved PV windows had significantly higher daylighting availability compared to flat windows,with the growth rates of the spatial useful daylight illuminance ranging of 3.94%–4.78%and 5.56%–5.94%,respectively,for the curved PV windows at central angles of 120°and 180°across different climate *** 120°curved PV windows achieved the lowest net energy used intensity(Net_EUI),suggesting the advantages of curved PV windows and proposed the existence of an optimal curvature for achieving the lowest Net_EUI.

关键词： BIPV curved PV PV window building performance

来源：评论

学校读者我要写书评

暂无评论

Target Detection and Adaptive Tracking Based on Multisensor Data Fusion in a Smoke Environment 8

Target Detection and Adaptive Tracking Based on Multisensor ...

引用

8th International Conference on Control, Automation and Robotics, ICCAR 2022

作者： Jiang, Chao Wang, Zhiling Liang, Huawei University Of Science And Technology Of China HeFei230026 China Anhui Engineering Laboratory For Intelligent Driving Technology And Application HeFei230061 China Innovation Research Institute Of Robotics And Intelligent Manufacturing Chinese Academy Of Sciences HeFei230061 China

ISBN: (纸本)9781665481168

Robust and accurate target detection and tracking in complex environments such as smoke are prerequisites for intelligent vehicle safe trajectory planning and navigation. Existing target detection and tracking methods have disadvantages, such as a high miss detection rate and apparent tracking deviation (position drift and size deviation of the tracking box) in a smoke environment. In response to these problems, we propose utilizing multisensor data fusion technology to improve the target detection recall and propose a new adaptive KCF tracking algorithm to improve tracking accuracy. We collected data from the environment of smoke to train the object detection network and designed a radar noise filter based on an empirical model to improve the accuracy and credibility of target detection. We then use a decision-level fusion strategy to fuse the image detection results and the radar filtering results. Finally, we utilize the fusion data to correct the tracking results online. The proposed method is evaluated on a self-developed intelligent vehicle platform, and the experimental results prove the algorithm's feasibility. This work provides a feasible solution to the problem of target detection and tracking in smoke environments. © 2022 IEEE.

关键词： Smoke

来源：评论

学校读者我要写书评

暂无评论

Real-Time Image Stitching with Transformers for Complex Traffic Environment 18th

Real-Time Image Stitching with Transformers for Complex Traf...

引用

18th Chinese Conference on Image and Graphics technology and application Conference, IGTA 2023

作者： Shen, Zhiwei Kong, Bin Institute of Intelligent Machines Chinese Academy of Sciences Hefei230031 China University of Science and Technology of China Hefei230026 China Anhui Engineering Laboratory for Intelligent Driving Technology and Application Hefei230088 China Chinese Academy of Sciences Hefei230088 China

ISBN: (纸本)9789819975488

The existing traditional image stitching methods suffer from accumulating errors in the matching process, which severely limits the final stitching results. Additionally, they cannot find correct correspondences in complex scenes with poor texture, repetitive patterns, and lighting variations. To address these issues, we propose a real-time image stitching method based on the Transformer structure. The self-attention mechanism of the Transformer can dynamically adjust the receptive field based on input content, allowing for effective global information retrieval. Compared to traditional manual and CNN-based deep learning methods, our approach generates dense and high-quality matches even in complex environments. We reduce the complexity of the Transformer model and improve matching speed through knowledge distillation. We use the Kmeans algorithm to optimize point selection for feature point screening and compute the homography matrix value based on the matching result for pixel transformation. After aligning the overlapping regions of the images and averaging them, we apply mask weights obtained from Gaussian fusion of target image layers to achieve the final stitching result. Experimental results demonstrate that our method produces more accurate matches in complex scenes with weak textures compared to traditional and deep learning methods, while also having the shortest matching and stitching times and good stitching results. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

Reference Path Generation for Autonomous driving in Off-road Environment 2022

Reference Path Generation for Autonomous Driving in Off-road...

引用

2nd International Conference on Robotics and Control engineering, RobCE 2022

作者： Zhang, Song Liang, Huawei Zhao, Pan Li, Zhiyuan Su, Tao Hefei230031 China University of Science and Technology of China Hefei230026 China Anhui Engineering Laboratory for Intelligent Driving Technology and Application Hefei230031 China Innovation Research Institute of Robotics and Intelligent Manufacturing Chinese Academy of Sciences Hefei230031 China

ISBN: (纸本)9781450395854

This paper proposes a reference path generation method for Autonomous Vehicles(AVs) in the off-road environment. Since obtaining sufficient and reliable prior information in an off-road environment is difficult, we adopt the Astar strategy to complete global routing based on the satellite topology map. Then planning result is corrected based on vehicle localization to obtain the initial reference path. Because the road geometry changes according to the road environment, we will use the centerline of the drivable area extracted with Probabilistic Neural Networks(PNN) to generate the final reference path. While the drivable area is invalid, the reference path planner is applied based on the initial reference path to form the final reference path. In the real off-road scenarios, the proposed method is tested on our experimental vehicle. Experimental results demonstrated that the proposed algorithm could generate reliable reference paths effectively and improve the performance and robustness of local trajectory planning. © 2022 Owner/Author.

关键词： Autonomous vehicles

来源：评论

学校读者我要写书评

暂无评论

Path Tracking Control Algorithm Considering Delay Compensation 7

Path Tracking Control Algorithm Considering Delay Compensati...

引用

7th Asia-Pacific Conference on intelligent Robot Systems, ACIRS 2022

作者： Zhang, Yujie Niu, Runxin Wang, Jie Liang, Huawei Chen, Zhengwei Huang, Zidong Hefei Institutes of Physical Science Chinese Academy of Sciences Hefei China University of Science and Technology of China Hefei China Anhui Engineering Laboratory for Intelligent Driving Technology and Application Heifei China Chinese Academy of Sciences Heifei China

ISBN: (数字)9781665485197

ISBN: (纸本)9781665485197

The delay in the control loop of autonomous vehicles can diminish controller effectiveness to a great extent and may even engender instability in a critical driving situation. Aiming at the path tracking problem with delays in the control loop, this paper presents a model predictive control method that considers the delay compensation. Firstly, the nonlinear vehicle dynamics model with delay is converted into a linear time-varying delay-free discrete system by expanding the steering delay command into the state variables of the system, and then the model is used as the predictive model of the model predictive control to design the controller. Finally, the proposed control system is evaluated by both co-simulation of MATLAB/Simulink & CarSim. Compared with the traditional model predictive control algorithm, the control compensation strategy based on the model predictive control algorithm proposed in this paper can significantly improve the robustness and stability of the system. © 2022 IEEE.

关键词： Model predictive control

来源：评论

学校读者我要写书评

暂无评论

Robust Curb Detection Based on the Accessible Route Analysis and Key Frames Prediction

Robust Curb Detection Based on the Accessible Route Analysis...

引用

2021 IEEE International Conference on Artificial Intelligence and Industrial Design, AIID 2021

作者： Yu, Zehai Zhu, Hui Lin, Linglong Yang, Haozhe Hefei Institutes of Physical Science Chinese Academy of Sciences Hefei230031 China University of Science and Technology of China Hefei230026 China Anhui Engineering Laboratory for Intelligent Driving Technology and Application Anhui Hefei230031 China Innovation Research Institute of Robotics and Intelligent Manufacturing Chinese Academy of Sciences Anhui Hefei230031 China

ISBN: (纸本)9780738110837

For autonomous driving in urban environments, road curb plays a significant role in tasks such as lane-keeping, assisted localization, and path planning. A real-time robust curb detection algorithm based on 3D LiDAR is proposed in this paper. Firstly, the iterative beam model is applied to get the accessible route of the road to obtain the starting point of the search step for each scan line. Secondly, the candidate curb points are extracted according to the spatial distribution characteristics of the point cloud. To effectively combine the historical boundaries information, a Bayesian filter is used to track the road width to reduce the false detection of curb points when the boundaries are interrupted, or on-road obstacles appear. The proposed algorithm is tested in different road environments. The experimental results show that our method has strong scene adaptability. The detection accuracy is over 90%, and the average runtime is 34.62 ms. © 2021 IEEE.

关键词： Autonomous vehicles

来源：评论

学校读者我要写书评

暂无评论

MAIM-VO: A Robust Visual Odometry with Mixed MLP for Weak Textured Environment 18th

MAIM-VO: A Robust Visual Odometry with Mixed MLP for Weak Te...

引用

18th Chinese Conference on Image and Graphics technology and application Conference, IGTA 2023

ISBN: (纸本)9789819975488

Visual localization is a critical technology for visual SLAM systems, which determines the relative position and motion trajectory by tracking feature points. In recent years, deep learning has been widely applied to the field of visual localization. The method based on deep learning is capable of surpassing the limitations of traditional manual feature extraction methods and achieving high-precision visual localization in complex scenes, thus realizing the goal of lifelong SLAM. The MLP model has characteristics such as flexibility and adaptability. The Mixer-WMLP achieves token information exchange between spatial positions by evenly dividing the feature map into non-overlapping windows, which makes the Mixer-WMLP approach a global receptive field. Compared to CNNs and Transformers, Mixer MLPs have higher computational efficiency and robustness. In this paper, we utilize the Mixer MLP structure to design a deep learning-based visual odometry system called MAIM-VO. Even in complex scenes with low texture areas, high-quality matching can be achieved. After obtaining the matching point pairs, the camera pose is solved in an optimized way by minimizing the reprojection error of the feature points. Multiple datasets and experiments in real-world environments have demonstrated that MIAM-VO exhibits higher robustness and relative localization accuracy compared to currently popular visual SLAM systems. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： Mixers (machinery)

来源：评论

学校读者我要写书评

暂无评论

Multi-feature fusion based region of interest generation method for far-infrared pedestrian detection system

Multi-feature fusion based region of interest generation met...

引用

2018 IEEE intelligent Vehicles Symposium, IV 2018

作者： Wang, Zhiling Lin, Linglong Li, Yuxin Chinese Academy of Sciences Anhui Engineering Laboratory for Intelligent Driving Technology and Appllication Hefei Institute of Physical Science Hefei Anhui230031 China Chinese Academy of Sciences Anhui Province Key Laboratory of Intelligent Connected Technology Hefei Institute of Physical Science Hefei Anhui230031 China

ISBN: (纸本)9781538644522

Given the uncontrolled outdoor environments and different physical properties of clothing, the appearance of pedestrians in far-infrared (FIR) images changes dramatically. Finding a robust region of interest (ROI) generation method for pedestrian detection remains challenging. Previous researches can obtain reliable results in some conditions. But they always got inappropriate results in warmer conditions. This study presents a Multi-feature fusion based ROI generation method for FIR pedestrian detection system to solve this problem. We extract two kinds of salient feature regions, namely, highlighting feature areas and vertical feature areas. A reasonable threshold is set to derive the highlighting feature areas and Scharr operator is used to find vertical edges. These areas are not necessarily connected to each other in the image. We think an upright pedestrian is a highly structured target consisting of highlighting feature areas and vertical feature areas. The distribution of feature areas is demonstrated using a skeleton model. So we apply the dilating morphological operation to ensure that these adjacent feature areas within pedestrians' regions will connect together. The size of the structuring element is set adaptively according to the size of feature areas. Finally, the experimental results show the robust performance of our method in different ambient conditions. © 2018 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：