检索结果-内蒙古大学图书馆

3D Scene Flow Estimation on Pseudo-LiDAR: Bridging the Gap on Estimating Point Motion

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Jiang, Chaokang Wang, Guangming Miao, Yanzi Wang, Hesheng Engineering Research Center of Intelligent Control for Underground Space Ministry of Education School of Information and Control Engineering Advanced Robotics Research Center China University of Mining and Technology Xuzhou221116 China Department of Automation Key Laboratory of System Control and Information Processing Ministry of Education Key Laboratory of Marine Intelligent Equipment and System of Ministry of Education Shanghai Engineering Research Center of Intelligent Control and Management Shanghai Jiao Tong University Shanghai200240 China

3D scene flow characterizes how the points at the current time flow to the next time in the 3D Euclidean space, which possesses the capacity to infer autonomously the non-rigid motion of all objects in the scene. The previous methods for estimating scene flow from images have limitations, which split the holistic nature of 3D scene flow by estimating optical flow and disparity separately. Learning 3D scene flow from point clouds also faces the difficulties of the gap between synthesized and real data and the sparsity of LiDAR point clouds. In this paper, the generated dense depth map is utilized to obtain explicit 3D coordinates, which achieves direct learning of 3D scene flow from 2D images. The stability of the predicted scene flow is improved by introducing the dense nature of 2D pixels into the 3D space. Outliers in the generated 3D point cloud are removed by statistical methods to weaken the impact of noisy points on the 3D scene flow estimation task. Disparity consistency loss is proposed to achieve more effective unsupervised learning of 3D scene flow. The proposed method of self-supervised learning of 3D scene flow on real-world images is compared with a variety of methods for learning on the synthesized dataset and learning on LiDAR point clouds. The comparisons of multiple scene flow metrics are shown to demonstrate the effectiveness and superiority of introducing pseudo-LiDAR point cloud to scene flow estimation. © 2022, CC BY.

关键词： Three dimensional computer graphics

Road Damage Detection and Classification Based on Multi-Scale Contextual Features

学校读者我要写书评

暂无评论

Road Damage Detection and Classification Based on Multi-Scal...

2022 IEEE International Conference on Big Data, Big Data 2022

作者： Yang, Lu He, Hualin Liu, Tao Tianjin University of Technology Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control School of Mechanical Engineering Tianjin300384 China Tianjin University of Technology National Demonstration Center for Experimental Mechanical and Electrical Engineering Education Tianjin300384 China

ISBN: (纸本)9781665480451

Road damage detection is particularly essential for improving road traffic safety. Existing methods do not perform well due to complex road conditions and sparse features contained in road damage. Therefore, we propose a new road damage detection network - YPLNet, which could obtain more refined multi-scale contextual features. Specifically, the network proposed in this paper is based on YOLOv5, by adding the PSA module to the backbone network, the spatial information of the feature map at multi-scale is obtained to enrich the feature space, and the long-term information dependence between channels is further constructed. In addition, in the feature fusion stage, the LCFI module is utilized to obtain contextual features under the multi-scale receptive fields, and these contextual features include rich low-level and high-level information. Our proposed network achieves a Mean-F1 Score of 0.635 in the competition. The source code and model are publicly available at https://***/hehualin-tut/YPLNet © 2022 IEEE.

关键词： Damage detection

Infinite-Horizon Optimal Wireless control Over Shared State-Dependent Fading Channels for IIoT systems

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wang, Shuling Li, Peizhe Zhu, Shanying Chen, Cailian Guan, Xinping Department of Automation Shanghai Jiao Tong University Shanghai200240 China Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai200240 China Shanghai Engineering Research Center of Intelligent Control and Management Shanghai200240 China

Heterogeneous systems consisting of a multiloop wireless control system (WCS) and a mobile agent system (MAS) are ubiquitous in Industrial Internet of Things systems. Within these systems, the positions of mobile agents may lead to shadow fading on the wireless channel that the WCS is controlled over and can significantly compromise its performance, requiring joint coordination between the WCS and MAS. Such coordination introduces different time steps and hybrid state spaces consisting of logical components and continuous components. This paper focuses on the infinite-horizon optimal control of MAS to ensure the performance of WCS while minimizing an average cost for the heterogeneous system subject to safety constraints. A state-dependent fading channel is modeled to capture interference among transmission links, as well as the effects of mobile agents’ movements on successful wireless transmission. In order to address the heterogeneous system dynamics, the optimal control problem is formulated as the optimal constrained set stabilization of the MAS by establishing a necessary and sufficient condition for the Lyapunov-like performance of WCS with the expected decay rates. Using the semi-tensor product of matrices, a constrained optimal state transition graph is constructed to encode the constrained system dynamics as well as objective function, which further reduces the problem into a minimum-mean cycle problem for the graph. By studying the properties of the graph, the feasibility is proven, and an effective algorithm is proposed for the construction of optimal input sequences. An illustrative example is provided to demonstrate effectiveness of the proposed method. Copyright © 2024, The Authors. All rights reserved.

关键词： Tensors

Active Target Tracking Using Bearing-only Measurements With Gaussian Process Learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Fu, Yingbo Yang, Ziwen Zhu, Shanying Guo, Yi Chen, Cailian The Department of Automation Shanghai Jiao Tong University Shanghai200240 China The Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai200240 China The Shanghai Engineering Research Center of Intelligent Control and Management Shanghai200240 China

This paper studies the tracking problem of a target with the partially unknown motion model by an active agent with bearing-only measurements using Gaussian process learning. To address this problem, a learning-planning-control framework is proposed. First, to learn and predict the target motion under mild assumptions, a Gaussian-process-based scheme is proposed, and a probabilistic uniform prediction error bound can be rigorously proved. Second, by analyzing the data dependence of the posterior covariance, we obtain an optimal relative trajectory to achieve efficient sampling. Third, to realize efficient learning, a controller to track the planned path is proposed based on the learned target motion, which can provide guaranteed tracking performance. Theoretical analysis is conducted to prove the the given probabilistic error bounds. Numerical examples and comparison with other typical methods verify the feasibility and superior performance of our proposed framework. Copyright © 2024, The Authors. All rights reserved.

关键词： Target tracking

Safety-Aware Scheduling of Real-Time control systems with Burst Computing Tasks

学校读者我要写书评

暂无评论

SSRN

SSRN 2024年

作者： Cheng, Ting Liang, Yonghui Xu, Qimin Zhu, Shanying Department of Automation Shanghai Jiao Tong University Shanghai200240 China Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai200240 China Shanghai Engineering Research Center of Intelligent Control and Management Shanghai200240 China

In industrial sites, multiple real-time control systems often share computing resources. However, burst computing tasks may lead to the dropping of control computing subtasks, resulting in potential hazards. To address this issue, we propose a safety-aware task scheduling scheme. When burst computing tasks are triggered, this scheduling scheme adaptively releases resources from low-criticality control computing subtasks, ensuring the timely completion of burst computing tasks and keeping the control system state deviation within a safe threshold. To achieve the scheduling scheme, we propose an efficient algorithm called ADP-Opt. It decouples the complex original problem into multiple subproblems, which greatly reduces the search space and improves the solution efficiency. In the suboptimization problem, safety is maintained by limiting the upper bound of the system state's deviation to below the threshold. By introducing a penalty term in the cost function related to computing resources, we aim to balance the impact of control computing subtasks on control performance and the consumption of computing resources. Consequently, the overall cost of the control system can be minimized while releasing enough resources for burst computing tasks. Simulation results show that the scheduling decisions made by ADP-Opt can approach the optimal decision by approximately 99% efficiently. © 2024, The Authors. All rights reserved.

关键词： Cost functions

Unsupervised Learning of 3D Scene Flow with 3D Odometry Assistance

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wang, Guangming Feng, Zhiheng Jiang, Chaokang Wang, Hesheng Department of Automation Key Laboratory of System Control and Information Processing of Ministry of Education Key Laboratory of Marine Intelligent Equipment and System of Ministry of Education Shanghai Engineering Research Center of Intelligent Control and Management Shanghai Jiao Tong University Shanghai200240 China Engineering Research Center of Intelligent Control for Underground Space Ministry of Education School of Information and Control Engineering Advanced Robotics Research Center China University of Mining and Technology Xuzhou221116 China

Scene flow represents the 3D motion of each point in the scene, which explicitly describes the distance and the direction of each point’s movement. Scene flow estimation is used in various applications such as autonomous driving fields, activity recognition, and virtual reality fields. As it is challenging to annotate scene flow with ground truth for real-world data, this leaves no real-world dataset available to provide a large amount of data with ground truth for scene flow estimation. Therefore, many works use synthesized data to pre-train their network and real-world LiDAR data to finetune. Unlike the previous unsupervised learning of scene flow in point clouds, we propose to use odometry information to assist the unsupervised learning of scene flow and use real-world LiDAR data to train our network. Supervised odometry provides more accurate shared cost volume for scene flow. In addition, the proposed network has mask-weighted warp layers to get a more accurate predicted point cloud. The warp operation means applying an estimated pose transformation or scene flow to a source point cloud to obtain a predicted point cloud and is the key to refining scene flow from coarse to fine. When performing warp operations, the points in different states use different weights for the pose transformation and scene flow transformation. We classify the states of points as static, dynamic, and occluded, where the static masks are used to divide static and dynamic points, and the occlusion masks are used to divide occluded points. The mask-weighted warp layer indicates that static masks and occlusion masks are used as weights when performing warp operations. Our designs are proved to be effective in ablation experiments. The experiment results show the promising prospect of an odometry-assisted unsupervised learning method for 3D scene flow in real-world data. Copyright © 2022, The Authors. All rights reserved.

关键词： Deep learning

Direct Visual Odometry by Fusing Luminosity and Depth Information

学校读者我要写书评

暂无评论

Direct Visual Odometry by Fusing Luminosity and Depth Inform...

2021 International Conference on Robotics Automation and intelligent control, ICRAIC 2021

作者： Yin, Jinming Zhou, Haibo Zhang, Haoxin Li, Chenming Sun, Guoqing Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control School of Mechanical Engineering Tianjin University of Technology Tianjin 300384 China National Demonstration Center for Experimental Mechanical and Electrical Engineering Education Tianjin University of Technology Tianjin 300384 China

In the traditional direct visual odometry, it is difficult to satisfy the photometric invariant assumption due to the influence of illumination changes in the real environment, which will lead to errors and drift. This paper proposes an improved direct visual odometry system, which combines luminosity and depth information. The algorithm proposed in this paper uses Kinect 2 to collect RGB images with the corresponding depth information, and selects points with large changes of gray gradient to construct a luminosity error function and uses the corresponding depth information to construct a depth error function. The two error functions are merged into one function and converted into the least squares function of the pose of camera, the Levenberg-Marquardt algorithm is used to solve the camera pose. Finally, the Graph optimization theory and the g2o library are used to optimize the initial pose. Experiments show that the algorithm can reduce the error to a certain extent and reduce the drift caused by illumination changes. © 2022 Institute of Physics Publishing. All rights reserved.

关键词： Vision

Unsupervised Learning of Gaussian Mixture Model with Application to Image Segmentation

学校读者我要写书评

暂无评论

Chinese Journal of Electronics 2023年第3期19卷 451-456页

作者： Bo Li Wenju Liu Lihua Dou School of Automation Beijing Institute of Technology Beijing China Key Laboratory of Complex System Intelligent Control and Decision Ministry of Education Beijing Institute of Technology Beijing China National Laboratory of Pattern Recognition Institute of Automation Chinese Academy of Sciences Beijing China

Density estimation via Gaussian mixture modeling has been successfully applied to image segmentation, speech processing and other fields relevant to clustering analysis and Probability density function (PDF) modeling. Finite Gaussian mixture model is usually used in practice and the selection of number of mixture components is a significant problem in its application. For example, in image segmentation, it is the donation of the number of segmentation regions. The determination of the optimal model order therefore is a problem that achieves widely attention. This paper proposes a degenerating model algorithm that could simultaneously select the optimal number of mixture components and estimate the parameters for Gaussian mixture model. Unlike traditional model order selection method, it does not need to select the optimal number of components from a set of candidate models. Based on the investigation on the property of the elliptically contoured distributions of generalized multivariate analysis, it select the correct model order in a different way that needs less operation times and less sensitive to the initial value of EM. The experimental results show the effectiveness of the algorithm.

关键词： Training Image segmentation Analytical models Parameter estimation Estimation Probability density function Speech processing

Virtual Sample Generation Methods and Application in Few-Shot Object Detection for Industrial Scenarios

学校读者我要写书评

暂无评论

Virtual Sample Generation Methods and Application in Few-Sho...

Chinese Automation Congress (CAC)

作者： Xiuwen Tao Guoyang Wan Ming Jiang Jian Zhang Qianqian Wang Qin He Key Laboratory of Advanced Perception and Intelligent Control of High-End Equipment Ministry of Education Anhui Polytechnic University School of Electrical Engineering Anhui Polytechnic University Wuhu China

Due to the long-term relatively fixed state of industrial installations, the diversity of industrial data is low, with high repetitiveness and uneven distribution, which makes industrial few-shot object detection still challenging. Therefore, from the perspective of handling the dataset, this paper proposes a virtual sample generation method based on virtual reality. Firstly, by using a virtual engine, the target workpiece model is created to directly simulate the states of the workpieces captured by the visual system in various environments. Secondly, random algorithms are added and improved to simulate workpieces with different materials and degrees of wear, distributed in random positions with different poses and types in the visual system, thereby generating virtual samples containing workpieces. Finally, virtual samples are introduced for experimental testing in the target detection network. The proposed method is tested in both YOLOv5 and YOLOv7 and compared with datasets without virtual samples, resulting in an improvement of 3.06% and 2.23% in mAP values, respectively. The experimental results demonstrate that the proposed method effectively improves the issue of industrial small-sample object detection.

关键词：