HWHaitao Wu

Haitao Wu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Haitao Wu has a total of 42 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    HOSG-Nav: Hierarchical Open-Vocabulary Semantic Graph Navigation for Language-Guided Global Planning in 3D Gaussian Scenes

    by Yuchen Li, Kai Qin, Weiyi Chen, Haitao Wu

    Natural-language-driven robot navigation in complex indoor environments requires the joint capability of high-fidelity scene representation, structured semantic reasoning, and executable path planning. To address this challenge, this paper proposes HOSG-Nav, a unified framework for natural-language-driven global navigation that integrates open-vocabulary 3D Gaussian scene representation, hierarchical semantic scene graph construction, and large-language-model-driven planning. First, an open-vocabulary 3D Gaussian field is constructed to jointly encode scene geometry, appearance, and semantic information, where compressed CLIP features are lifted into continuous 3D space and depth supervision is introduced to enhance geometric stability and metric-scale consistency. Second, the optimized Gaussian primitives are further abstracted into a semantic scene graph with a region–object hierarchical structure and traversable topological relations to support structured environment understanding. Finally, for natural language instructions, hierarchical semantic parsing is performed with the assistance of a large language model, and executable global navigation paths are generated through cross-modal target retrieval and graph-search-based planning. Experimental results on the Replica dataset demonstrate that HOSG-Nav achieves competitive performance in scene representation, semantic target retrieval, and global navigation, validating the effectiveness of jointly integrating multimodal 3D representation, hierarchical semantic abstraction, and language-guided planning.


  • Open Access

    Cross-Modal 3-D Gaussian Reconstruction: Fusing Image, Geometry, and Open-Vocabulary Semantics for Large-Scale Real-World Scenes

    by Yuchen Li, Jing Li, Kai Qin, Haitao Wu, Yin Gao, Banghui Yang

    3D Gaussian Splatting (3D GS) has revolutionized scene representation, efficient photorealistic rendering, and novel view synthesis. However, when applied to complex large-scale real-world scenes, 3D GS faces two major challenges which are geometric errors and rendering artifacts. To address these challenges, we propose FuseGS, a novel cross-modal Gaussian reconstruction framework that seamlessly fuses multi-view images and dense LiDAR point clouds, along with open-vocabulary semantics from large Vision-Language Models (VLMs) to achieve more accurate global geometry and realistic renderings than the existing methods. FuseGS introduces three key innovations. First, to overcome low geometric accuracy in image-only methods, we propose a novel LiDAR-guided Gaussian initialization method to ensure strong geometric supervision. Second, to deal with the difficulty of integrating 2D VLMs semantics into 3D space, we propose a novel category-specific dual-cue refinement method to improve semantics and geometry alignment. Third, to remove rendering artifacts caused by dynamic objects, we introduce a novel semantic-based adaptive density control method to effectively maintain static background reconstruction. To demonstrate the robustness of our framework, we evaluate FuseGS using both self-collected data from handheld and Unmanned Aerial Vehicle (UAV) platforms, as well as public datasets. Compared to the baseline 3D GS method, FuseGS unifies joint optimization of photometric, geometric and static-aware semantics, resulting in 50% improvement in geometric accuracy and 4.8 dB improvement in rendering quality. The method achieves 67.43% mean Intersection over Union (mIoU) without manual annotation. Ultimately, FuseGS has demonstrated excellent potential for Earth observation applications such as smart cities and autonomous driving.

  • Open Access

    ProM-Pose: Language-Guided Zero-Shot 9-DoF Object Pose Estimation from RGB-D with Generative 3D Priors

    by Yuchen Li, Kai Qin, Haitao Wu, Xiangjun Qu

    Object pose estimation is fundamental for robotic manipulation, autonomous driving, and augmented reality, yet recovering the full 9-DoF state (rotation, translation, and anisotropic 3D scale) from RGB-D observations remains challenging for previously unseen objects. Existing methods either rely on instance-specific CAD models, predefined category boundaries, or suffer from scale ambiguity under sparse observations. We propose ProM-Pose, a unified cross-modal temporal perception framework for zero-shot 9-DoF object pose estimation. By integrating language-conditioned generative 3D shape priors as canonical geometric references, an asymmetric cross-modal attention mechanism for spatially aware fusion, and a decoupled pose decoding strategy with temporal refinement, ProM-Pose constructs metrically consistent and semantically grounded representations without relying on category-specific pose priors or instance-level CAD supervision. Extensive experiments on CAMERA25 and REAL275 benchmarks demonstrate that ProM-Pose achieves competitive or superior performance compared to category-level methods, with mAP of 75.0% at 5°,2cm and 90.5% at 10°,5cm on CAMERA25, and 42.2% at 5°,2cm and 76.0% at 10°,5cm on REAL275 under zero-shot cross-domain evaluation. Qualitative results on real-world logistics scenarios further validate temporal stability and robustness under occlusion and lighting variations. ProM-Pose effectively bridges semantic grounding and metric geometric reasoning within a unified formulation, enabling stable and scale-aware 9-DoF pose estimation for previously unseen objects under open-vocabulary conditions.


  • Open Access

    VGGT-Geo: Probabilistic Geometric Fusion of Visual Geometry Grounded Transformer Priors for Robust Dense Indoor SLAM

    by Kai Qin, Jing Li, Sisi Zlatanova, Haitao Wu, Hao Wu, Yin Gao, Dingjie Zhou, Yuchen Li, Sizhe Shen, Xiangjun Qu, et al.

    With the rapid evolution of Digital Twins and Embodied AI, achieving fast, dense, and high-precision 3D perception in unknown environments has become paramount. However, existing Visual SLAM paradigms face a critical dilemma: geometry-based methods often fail in texture-less areas due to feature scarcity, while learning-based approaches frequently suffer from scale drift and unphysical deformations. To bridge this gap, we propose VGGT-Geo, a novel SLAM system that synergizes generative priors from Large Foundation Models with multi-modal geometric optimization. Distinguishing itself from simple cascaded architectures, we construct a Probabilistic Geometric Fusion framework, consisting of (1) Generative Warm-start, leveraging the holistic scene understanding capabilities of the VGGT, (2) Confidence-Aware Optimization to extract dense features via DINOv3 and predict their confidence map, and (3) a Multi-Modal Constraint Closure that fuses point-line features and metric depth priors to constrain rotational Degrees of Freedom in Manhattan Worlds. We conducted systematic evaluations on TUM, Replica, Tanks and Temples, and a challenging self-collected dataset featuring extreme lighting and texture-less walls. Experimental results demonstrate that VGGT-Geo exhibits superior robustness and accuracy in unseen environments. On our most challenging dataset, it achieves an Absolute Trajectory Error of 4–5 cm and a Relative Rotation Error of 0.79°, outperforming current state-of-the-art methods by approximately 50% in trajectory accuracy. This study validates that synergizing the intuition of Large Foundation Models with geometric rigor is a viable path toward next-generation robust SLAM.


  • Attitude-Independent Magnetic Vector Positioning System With Multilayer Cascaded Attenuation Modeling and Frequency Conversion Calibration

    by Guan Wang, Wenchao Zhang, Dongyan Wei, Hong Yuan, Haitao Wu

    Artificial magnetic field positioning shows promise for indoor localization by overcoming physical obstructions, yet it suffers from complex environmental interference and unknown attitudes orientation ambiguity affecting accuracy. To address the accuracy limitation, this paper proposes an attitude-independent magnetic vector positioning system integrating multilayer cascaded attenuation modeling and frequency conversion calibration. Firstly, an air-building-air multilayer cascaded attenuation model is established based on the intra-medium magnetic diffusion equation and inter-medium propagation continuity theory. Secondly, a frequency conversion calibration method is proposed to address the interference variability arising from differences in building properties and spatial locations. The method exploits the frequency-dependent attenuation characteristics of multi-frequency magnetic signals to construct an adaptive model for responsive parameter estimation. Furthermore, an enhanced magnetic vector positioning algorithm is developed, formulating attitude-independent overdetermined observation equations by fusing magnitude scalar and vector information, with least-squares optimization for position determination. Field tests in a 128 m² industrial plant and a 700 m² multi-story office building demonstrate mean 2D and 3D positioning errors of 0.81 m and 0.95 m, respectively, achieving accuracy improvements of up to 49.7% over existing methods without positional ambiguity.

  • Multiepoch Least-Squares Joint Solution for Anchor Self-Localization: Exemplified With UWB

    by Lelong Zhao, Xiaochen Guo, Haitao Wu, Jian Ge

    In indoor positioning through the Internet of Things (IoT), the accuracy of anchor locations profoundly impacts receiver localization. At present, the determination of anchor positions primarily relies on traditional manual surveying techniques. The considerable costs of manual surveying have led the academic community to advocate for anchor self-localization methods as viable alternatives. However, existing methodologies prove inadequate in complex indoor environments lacking supplementary information sources, particularly in addressing non-line-of-sight (NLOS) errors that impact anchor positioning accuracy. To address these challenges, this study utilizes ultra-wideband (UWB) signal sources as a case study. We mainly propose a novel anchor self-localization method that relies solely on ranging measurements. The method integrates the proposed NLOS bias model based on temporal and spatial correlation to overcome NLOS effects. The method initially employs an initialization algorithm to obtain coarse estimates of anchor positions through multiepoch least squares (LS) joint estimation, followed by a two-step iterative algorithm that utilizes the expectation maximization principle to optimize these estimates further. The effectiveness of the proposed method is validated through anchor self-localization tests in indoor environments with decimeter-level positioning accuracy results. The root mean square error (RMSE) of the subsequent receiver positioning results was 0.1936 m, validating that the attained anchor positions satisfy pragmatic application standards.


  • Open Access

    Semantic Object-level Modeling for Robust Visual Camera Relocalization

    by Yifan Zhu, Lingjuan Miao, Haitao Wu, Zhiqiang Zhou, Weiyi Chen, Longwen Wu

    Visual relocalization is crucial for autonomous visual localization and navigation of mobile robotics. Due to the improvement of CNN-based object detection algorithm, the robustness of visual relocalization is greatly enhanced especially in viewpoints where classical methods fail. However, ellipsoids (quadrics) generated by axis-aligned object detection may limit the accuracy of the object-level representation and degenerate the performance of visual relocalization system. In this paper, we propose a novel method of automatic object-level voxel modeling for accurate ellipsoidal representations of objects. As for visual relocalization, we design a better pose optimization strategy for camera pose recovery, to fully utilize the projection characteristics of 2D fitted ellipses and the 3D accurate ellipsoids. All of these modules are entirely intergrated into visual SLAM system. Experimental results show that our semantic object-level mapping and object-based visual relocalization methods significantly enhance the performance of visual relocalization in terms of robustness to new viewpoints.

  • Framework and Methods of State Monitoring-Based Positioning System on WIFI-RTT Clock Drift Theory

    by Xiaochen Guo, Haitao Wu

    High-precision indoor positioning problems have attracted considerable attention recently. The indoor ranging and positioning method based on WIFI round-trip-time (RTT) has become popular. However, resulting from clock drift (CD), there are two problems that are still not completely resolved and need to be addressed. The first problem is ranging error drift, namely the reset of phase distortion errors (PDEs), while the second problem is the communication state change, namely ranging information loss and restoration. To solve these two problems, this study noticed that the reset of PDEs will only change along with the changes in communication states. Thus, a monitoring-based positioning system (MPS) framework is presented based on the theoretical model of clock drift as a method to account for ranging error drift and communication state change when estimating position with IEEE 802.11mc protocol. Then, a state monitoring algorithm (SMA) suitable for the MPS framework is proposed to accurately capture the moment of communication state change. Further, a PDE constraint (PDEC) model is proposed based on the SMA to estimate PDEs in different communication states to reduce the impact of CD on positioning results. The experimental results finally show that considerable improvement in positioning accuracy can be obtained using the SMA and PDEC model compared with that obtained using the semi-parametric solution algorithm


  • Open Access

    Satellite Navigation Signal Authentication in GNSS: A Survey on Technology Evolution, Status, and Perspective for BDS

    by Xiao Chen, Ruidan Luo, Ting Liu, Hong Yuan, Haitao Wu

    As the Global Navigation Satellite System (GNSS) is widely used in all walks of life, the signal structure of satellite navigation is open, and the vulnerability to spoofing attacks is also becoming increasingly prominent, which will seriously affect the credibility of navigation, positioning, and timing (PNT) services. Satellite navigation signal authentication technology is an emerging technical means of improving civil signal anti-spoofing on the satellite navigation system side, and it is also an important development direction and research focus of the GNSS. China plans to carry out the design and development of the next-generation Beidou navigation satellite system (BDS), and one of its core goals is to provide more secure and credible PNT services. This paper first expounds on the principles and technical architecture of satellite navigation signal authentication, then clarifies the development history of satellite navigation signal authentication, and finally proposes the BDS authentication service system architecture. It will provide technical support for the construction and development of the follow-up Beidou authentication service.