Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Yipeng Ning has a total of 61 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Robust RGB-D Visual SLAM for Dynamic Environments via Image Deblurring and 3D Motion Estimation

    by Jinlong Fan, Yipeng Ning, Shengfeng Gu

    Visual simultaneous localization and mapping (V-SLAM) plays a crucial role in autonomous navigation, especially in Internet of Things (IoT) application scenarios. However, dynamic objects and rapid camera motion often introduce image blur and feature matching errors, thereby reducing the accuracy of pose estimation. To address these problems, we propose BD-SLAM, a robust and real-time RGB-D visual SLAM framework. Different from existing dynamic SLAM frameworks, we construct an adaptive image deblurring filter to restore image clarity, thereby providing reliable visual input. In addition, we propose a geometric constraint method based on 3D motion field estimation, which overcomes the limitations of relying solely on semantic priors and improves the accuracy of dynamic feature rejection. Furthermore, we design a semantic-guided dynamic point suppression algorithm, which preserves reliable static features while significantly reducing the computational overhead of geometric constraint. The proposed system is evaluated on the TUM, Bonn, and self-collected RGB-D datasets. Experimental results demonstrate that BD-SLAM achieves higher localization accuracy and robustness in dynamic environments compared with state-of-the-art SLAM methods.

  • Enhancing Dynamic Visual SLAM Precision Based on Pseudo-Semantic Segmentation and Features Weighting

    by Jinlong Fan, Yipeng Ning, Zhengwei Guo, Xiaoming Xi, Dashuai Chai, Wengang Sang, Feng Zhu

    Localization accuracy in dynamic Simultaneous Localization and Mapping (SLAM) is degraded when moving objects violate the static-environment assumption. This paper proposes a dynamic Visual SLAM (VSLAM) method that integrates pseudo-semantic segmentation with weighted optimization to address the issue. First, an adaptive feature extraction module is designed, which establishes a Gaussian distribution model for intra-frame features and combines it with an inter-frame matching quality function to overcome the limitations of traditional feature extraction methods. Second, a depth-based pseudo-semantic segmentation method is proposed, where the K-means clustering algorithm is employed to perform clustering analysis on depth information, generating accurate dynamic object masks for precise exclusion of dynamic features. Finally, a weighted optimization model based on epipolar constraints is constructed. By calculating the distance from feature points to epipolar lines and assigning weights accordingly, the model suppresses interference from dynamic feature points while enhancing the contribution of high-confidence static feature points. Experimental results demonstrate that, compared to ORB-SLAM3, the proposed method reduces the root mean square error (RMSE) of the Absolute Trajectory Error (ATE) by approximately 98.5% on the TUM dataset, while outperforming other existing state-of-the-art methods in terms of localization accuracy and robustness.

  • Open Access

    YSAG-VINS—A Robust Visual-Inertial Navigation System with Adaptive Geometric Constraints and Semantic Information Based on YOLOv8n-ODUIB in Dynamic Environments

    by Kunlin Wang, Dashuai Chai, Xiqi Wang, Ruijie Yan, Yipeng Ning, Wengang Sang, Shengli Wang

    Dynamic environments pose significant challenges for Visual Simultaneous Localization and Mapping (VSLAM), as moving objects can introduce outlier observations that severely degrade localization and mapping performance. To address this problem, we propose YSAG-VINS, a VSLAM algorithm specifically designed for dynamic scenes. The system integrates an enhanced YOLOv8 object detection network with an adaptive epipolar constraint strategy to effectively identify and suppress the impact of dynamic features. In particular, a lightweight YOLOv8n model augmented with ODConv and UIB modules is employed to balance detection accuracy with real-time efficiency. Based on semantic detection results, images are divided into static background and potentially dynamic regions, and the motion state of these regions is further verified using geometric constraints. Features belonging to truly dynamic objects are then removed to enhance robustness. Comprehensive experiments on multiple public datasets demonstrate that YSAG-VINS achieves superior pose estimation accuracy compared with VINS-Fusion, VDO-SLAM, and Dynamic-VINS. On three dynamic sequences of the KITTI dataset, the proposed method achieves average RMSE improvement rates of 48.62%, 12.18%, and 13.50%, respectively. These results confirm that YSAG-VINS provides robust and high-accuracy localization performance in dynamic environments, making it a promising solution for real-world applications such as autonomous driving, service robotics, and augmented reality.


  • Prior Distribution Guided Gaussian Mixture Variational Autoencoder (PDGM-VAE) for Image Generation

    by Jingqi Song, Zhiqiang He, Yipeng Ning, Xiaoming Xi, Jie Guo, Guanzhong Chen, Xiushan Nie, Lishan Qiao, Yilong Yin

    Variational Autoencoder(VAE) combines the ideas of autoencoders and variational inference, introducing the concept of latent space and variational inference to endow autoencoders to generate new images. VAE typically assumes that data follows a Gaussian distribution, but real data may follow other distributions. This inconsistency between the assumption and the true distribution can affect the modeling and reconstruction capabilities of VAE, which makes it difficult for traditional models to accurately capture the true distribution. To address the aforementioned issues, we propose a Prior Distribution Guided Gaussian Mixture Variational Autoencoder(PDGM-VAE). Specifically, we construct a Gaussian Mixture Prior Learner (GMPL) to capture complex features of the data distribution, enabling the model to learn and obtain a Gaussian mixture distribution that is reasonable and close to the real data distributions, which is then used as the prior distribution in the network. Furthermore, we build a Semantic-Aware Module with Embedded Prior Distribution (SAMEPD), integrating data and label information to learn the distribution parameters, enabling the network to learn and utilize the semantic knowledge contained in the labels. During training, by approximating the posterior distribution to the prior distribution, we enhance the model’s modeling and reconstruction capabilities, improving the quality of generated images. We evaluated the image generation task on five public datasets, and based on the FID metric, our proposed method outperformed other VAE methods.


  • Open Access

    Partial Ambiguity Resolution Strategy for Single-Frequency GNSS RTK/INS Tightly Coupled Integration in Urban Environments

    by Dashuai Chai, Xiqi Wang, Yipeng Ning, Wengang Sang

    Single-frequency global navigation satellite system/inertial navigation system (GNSS/INS) integration has wide application prospects in urban environments; however, correct integer ambiguity is the major challenge because of GNSS-blocked environments. In this paper, a sequential strategy of partial ambiguity resolution (PAR) of GNSS/INS for tightly coupled integration based on the robust posteriori residual, elevation angle, and azimuth in the body frame using INS aids is presented. First, the satellite is eliminated if the maximum absolute value of the robust posteriori residuals exceeds the set threshold. Otherwise, the satellites with a minimum elevation angle of less than or equal to 35° are successively eliminated. If satellites have elevation angles greater than 35°, these satellites are divided into different quadrants based on their azimuths calculated in body frame. The satellite with the maximum azimuth in each quadrant is selected as the candidate satellite, the candidate satellites are eliminated one by one, and the remaining satellites are used to calculate the position dilution of the precision (PDOP). Finally, the candidate satellite with the lowest PDOP is eliminated. Two sets of vehicle-borne data with a low-cost GNSS/INS integrated system are used to analyze the performance of the proposed algorithm. These experiments demonstrate that the proposed algorithm has the highest ambiguity fixing rates among all the designed PAR methods, and the fixing rates for these two sets of data are 99.40% and 98.74%, respectively. Additionally, among all the methods compared in this paper, the proposed algorithm demonstrates the best positioning performance in GNSS-blocked environments.


  • Open Access

    EMS-SLAM: Dynamic RGB-D SLAM with Semantic-Geometric Constraints for GNSS-Denied Environments

    by Jinlong Fan, Yipeng Ning, Jian Wang, Xiang Jia, Dashuai Chai, Xiqi Wang, Ying Xu

    Global navigation satellite systems (GNSSs) exhibit significant performance limitations in signal-deprived environments such as indoor spaces and underground spaces. Although visual SLAM has emerged as a viable solution for ego-motion estimation in GNSS-denied areas, conventional approaches remain constrained by static environment assumptions, resulting in a substantial degradation in accuracy when handling dynamic scenarios. The EMS-SLAM framework combines the geometric constraints and semantics of SLAM to provide a real-time solution for addressing the challenges of robustness and accuracy in dynamic environments. To improve the accuracy of the initial pose, EMS-SLAM employs a feature-matching algorithm based on a graph-cut RANSAC. In addition, a degeneracy-resistant geometric constraint method is proposed, which effectively addresses the degeneracy issues of purely epipolar approaches. Finally, EMS-SLAM combines semantic information with geometric constraints to maintain high accuracy while quickly eliminating dynamic feature points. Experiments were conducted on the public datasets and our collected datasets. The results demonstrate that our method outperformed the current algorithms of SLAM in highly dynamic environments.


  • Open Access

    Optimization Model-Based Robust Method and Performance Evaluation of GNSS/INS Integrated Navigation for Urban Scenes

    by Dashuai Chai, Shijie Song, Kunlin Wang, Jingxue Bi, Yunlong Zhang, Yipeng Ning, Ruijie Yan

    The robust and high-precision estimation of position and attitude information using a combined global navigation satellite system/inertial navigation system (GNSS/INS) model is essential to a wide range of applications in intelligent driving and smart transportation. GNSS systems are susceptible to inaccuracies and signal interruptions in occluded environments, which lead to unreliable parameter estimations in GNSS/INS based on filter models. To address this issue, in this paper, a GNSS/INS combination model based on factor graph optimization (FGO) is investigated and the robustness of this optimization model is evaluated in comparison to the traditional extended Kalman filter (EKF) model and robust Kalman filter (RKF) model. In this paper, both high- and low-accuracy GNSS/INS combination data are used and the two sets of urban scene data are collected using high- and low-precision consumer-grade inertial guidance systems and an in-vehicle setup. The experimental results demonstrate that the position, velocity, and attitude estimates obtained using the GNSS/INS and the FGO model are superior to those obtained using the traditional EKF and robust EKF methods. In the simulated scenarios involving gross interference and GNSS signal loss, the FGO model achieves optimal results. The maximum improvement rates of the position, velocity, and attitude estimates are 81.1%, 73.8%, and 75.1% compared to the EKF method and 79.8%, 72.1%, and 57.1% compared to the RKF method, respectively.


  • EGLT-SLAM: Real-Time Visual-Inertial SLAM Based on Entropy-Guided Line Tracking

    by Xiang Jia, Yipeng Ning, Dashuai Chai, Jinlong Fan, Zhen Yang, Xiaoming Xi, Feng Zhu, Weiwei Wang

    Visual Simultaneous Localization and Mapping (SLAM) systems frequently encounter challenges in low-texture environments, resulting in suboptimal performance. Integrating line features into SLAM has emerged as an effective solution. However, a significant drawback of line feature-based SLAM is its limited capability for real-time data processing. Hence, we present EGLT-SLAM, a real-time point-line fusion visual-inertial SLAM system. To promote the system’s robustness in complex environments, the traditional line extraction algorithm is enhanced to assist the feature tracking. An optical flow tracking model specifically designed for line features is developed to further optimize the real-time performance of the SLAM system. Our approach involves extracting line segments from individual frames and tracking them across multiple frames. Decisions for line segment extraction are based on the image entropy change threshold of specified keyframes within the sliding window. Furthermore, a line reprojection error metric that is independent of the endpoints of the line segments is introduced. Experimental results demonstrate that accurate and real-time localization is achieved by EGLT-SLAM even in challenging scenarios with weak textures, illumination changes, and motion blur.


  • Open Access

    Research on a Matching Method for Vehicle-Borne Laser Point Cloud and Panoramic Images Based on Occlusion Removal

    by Jiashu Ji, Weiwei Wang, Yipeng Ning, Hanwen Bo, Yufei Ren

    Vehicle-borne mobile mapping systems (MMSs) have been proven as an efficient means of photogrammetry and remote sensing, as they simultaneously acquire panoramic images, point clouds, and positional information along the collection route from a ground-based perspective. Obtaining accurate matching results between point clouds and images is a key issue in data application from vehicle-borne MMSs. Traditional matching methods, such as point cloud projection, depth map generation, and point cloud coloring, are significantly affected by the processing methods of point clouds and matching logic. In this study, we propose a method for generating matching relationships based on panoramic images, utilizing the raw point cloud map, a series of trajectory points, and the corresponding panoramic images acquired using a vehicle-borne MMS as input data. Through a point-cloud-processing workflow, irrelevant points in the point cloud map are removed, and the point cloud scenes corresponding to the trajectory points are extracted. A collinear model based on spherical projection is employed during the matching process to project the point cloud scenes to the panoramic images. An algorithm for vectorial angle selection is also designed to address filtering out the occluded point cloud projections during the matching process, generating a series of matching results between point clouds and panoramic images corresponding to the trajectory points. Experimental verification indicates that the method generates matching results with an average pixel error of approximately 2.82 pixels, and an average positional error of approximately 4 cm, thus demonstrating efficient processing. This method is suitable for the data fusion of panoramic images and point clouds acquired using vehicle-borne MMSs in road scenes, provides support for various algorithms based on visual features, and has promising applications in fields such as navigation, positioning, surveying, and mapping.