Xiao Huang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xiao Huang has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Jing Xu
Hubei Key Laboratory of Internet of Intelligence, School of Electronics Information and Communications, Huazhong University of Science and Technology, Wuhan, China
4 shared publications - Wei Liu
Huazhong University of Science and Technology
4 shared publications - Jie Ma
Huazhong University of Science and Technology
3 shared publications - Jiarun Tang
Hubei Key Laboratory of Internet of Intelligence, School of Electronic Information and Communications, Huazhong University of Science and Technology, Wuhan, China
2 shared publications - Jie Ma
School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, Wuhan, China
2 shared publications - Junfeng Ding
Huazhong University of Science and Technology
2 shared publications
Latest Publications
- 10 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (6) , 11454-11466
Encoder-Decoder Based Deep Reinforcement Learning for Multi-AUV Assisted Data Collection in UWSNs
Reliable and timely data collection poses a significant challenge for underwater wireless sensor networks (UWSNs), primarily due to the extremely low data rate of underwater communication and the difficulty in supplying energy to the sensor nodes. In this paper, we explore the use of multiple autonomous underwater vehicles (AUVs) for sequential data collection from sensor nodes distributed over a large area. To facilitate efficient data collection, appropriate cluster heads (CHs) are selected in each region to collect and aggregate data from nearby energy-limited sensor nodes, and then forward the data to the mobile AUVs. Specifically, we investigate how to maximize the value of information (VoI) of the collected data through the joint optimization of CH selection and multi-AUV path planning, while accounting for the energy constraints of sensor nodes. To tackle the formulated VoI maximization problem, we propose a deep reinforcement learning algorithm based on an encoder-decoder architecture. The whole UWSN system is fed into the encoder network, followed by a composite decoder consisting of an AUV selection decoder and a cluster access sequence decoder to generate the cluster access sequence for each AUV. Based on the determined sequences, we further utilize the dynamic programming algorithm to achieve optimal CH selection. Finally, we obtain the sequence of AUVs that access the selected CHs. Simulation results demonstrate that the proposed learning-based approach converges and achieves a higher VoI than existing benchmark algorithms.
- 10 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (6) , 11454-11466
Encoder-Decoder Based Deep Reinforcement Learning for Multi-AUV Assisted Data Collection in UWSNs
Reliable and timely data collection poses a significant challenge for underwater wireless sensor networks (UWSNs), primarily due to the extremely low data rate of underwater communication and the difficulty in supplying energy to the sensor nodes. In this paper, we explore the use of multiple autonomous underwater vehicles (AUVs) for sequential data collection from sensor nodes distributed over a large area. To facilitate efficient data collection, appropriate cluster heads (CHs) are selected in each region to collect and aggregate data from nearby energy-limited sensor nodes, and then forward the data to the mobile AUVs. Specifically, we investigate how to maximize the value of information (VoI) of the collected data through the joint optimization of CH selection and multi-AUV path planning, while accounting for the energy constraints of sensor nodes. To tackle the formulated VoI maximization problem, we propose a deep reinforcement learning algorithm based on an encoder-decoder architecture. The whole UWSN system is fed into the encoder network, followed by a composite decoder consisting of an AUV selection decoder and a cluster access sequence decoder to generate the cluster access sequence for each AUV. Based on the determined sequences, we further utilize the dynamic programming algorithm to achieve optimal CH selection. Finally, we obtain the sequence of AUVs that access the selected CHs. Simulation results demonstrate that the proposed learning-based approach converges and achieves a higher VoI than existing benchmark algorithms.
- 23 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (15) , 29953-29965
Enhancing Underwater Wireless Sensor Network Lifetime: An Energy-Aware AUV-Assisted Data Collection Scheme
Underwater wireless sensor networks (UWSNs) play a crucial role in various applications, including marine environmental monitoring, underwater resource exploration, and intrusion detection. Autonomous underwater vehicle (AUV)-assisted data collection has emerged as a highly promising approach to enhance the capabilities of UWSNs. However, the operation of these networks faces significant challenges, primarily due to the limited and non-replaceable battery power of sensor nodes and the energy constraint of the AUV. To address these challenges and improve the network lifetime of UWSNs, this paper proposes an energy-aware AUV-assisted data collection scheme, named DCCHS, which combines dynamic clustering with cluster head (CH) selection. The DCCHS scheme employs a bottom-up hierarchical clustering algorithm based on the energy center to adaptively form clusters under the constraint of AUV movement distance. Furthermore, an iterative optimization algorithm is applied to select CHs based on residual energy. To balance energy consumption and prevent premature network depletion, auxiliary CHs are strategically added near the AUV path. Simulation results demonstrate that the proposed DCCHS scheme significantly outperforms existing methods in extending network lifetime, particularly in scenarios with dense node deployment. This improvement is attributed to the effective balancing of energy consumption among sensor nodes and the AUV.
- 23 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (15) , 29953-29965
Enhancing Underwater Wireless Sensor Network Lifetime: An Energy-Aware AUV-Assisted Data Collection Scheme
Underwater wireless sensor networks (UWSNs) play a crucial role in various applications, including marine environmental monitoring, underwater resource exploration, and intrusion detection. Autonomous underwater vehicle (AUV)-assisted data collection has emerged as a highly promising approach to enhance the capabilities of UWSNs. However, the operation of these networks faces significant challenges, primarily due to the limited and non-replaceable battery power of sensor nodes and the energy constraint of the AUV. To address these challenges and improve the network lifetime of UWSNs, this paper proposes an energy-aware AUV-assisted data collection scheme, named DCCHS, which combines dynamic clustering with cluster head (CH) selection. The DCCHS scheme employs a bottom-up hierarchical clustering algorithm based on the energy center to adaptively form clusters under the constraint of AUV movement distance. Furthermore, an iterative optimization algorithm is applied to select CHs based on residual energy. To balance energy consumption and prevent premature network depletion, auxiliary CHs are strategically added near the AUV path. Simulation results demonstrate that the proposed DCCHS scheme significantly outperforms existing methods in extending network lifetime, particularly in scenarios with dense node deployment. This improvement is attributed to the effective balancing of energy consumption among sensor nodes and the AUV.
- 23 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (1) , 627-634
A Robust LiDAR-Camera Self-Calibration Via Rotation-Based Alignment and Multi-Level Cost Volume
Multi-sensor collaborative perception has been a significant trend in self-driving and robot navigation. The precondition for multi-sensor fusion is the accurate calibration between sensors. Traditional LiDAR-Camera calibrations rely on laborious manual operations. Several recent studies have demonstrated the advantages of convolutional neural networks regarding feature extraction capabilities. However, the vast modality discrepancy between RGB images and point clouds makes it difficult to explore corresponding features, remaining a challenge for LiDAR-Camera calibrations. In this letter, we propose a new robust online LiDAR-Camera self-calibration network (SCNet). To reduce the search dimensionality for feature matching, we exploit self-supervised learning to align RGB images with projected depth images in 2D pixel coordinates, thereby achieving pre-alignment of the roll angle. In addition, to generate more accurate initial similarity measures for RGB image pixels and possible corresponding projected depth image pixels, we propose a novel multi-level patch matching method that concatenates cost volume constructed from multi-level feature maps. Our method achieves a mean absolute calibration error of 0.724 cm in translation and 0.055 $^{\circ }$ in rotation in a single frame analysis with miscalibration magnitudes of up to $\pm$ 1.5 m and $\pm 20^{\circ }$ on the KITTI odometry dataset, which demonstrates the superiority of our method.
- 23 November 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 9 (1) , 627-634
A Robust LiDAR-Camera Self-Calibration Via Rotation-Based Alignment and Multi-Level Cost Volume
Multi-sensor collaborative perception has been a significant trend in self-driving and robot navigation. The precondition for multi-sensor fusion is the accurate calibration between sensors. Traditional LiDAR-Camera calibrations rely on laborious manual operations. Several recent studies have demonstrated the advantages of convolutional neural networks regarding feature extraction capabilities. However, the vast modality discrepancy between RGB images and point clouds makes it difficult to explore corresponding features, remaining a challenge for LiDAR-Camera calibrations. In this letter, we propose a new robust online LiDAR-Camera self-calibration network (SCNet). To reduce the search dimensionality for feature matching, we exploit self-supervised learning to align RGB images with projected depth images in 2D pixel coordinates, thereby achieving pre-alignment of the roll angle. In addition, to generate more accurate initial similarity measures for RGB image pixels and possible corresponding projected depth image pixels, we propose a novel multi-level patch matching method that concatenates cost volume constructed from multi-level feature maps. Our method achieves a mean absolute calibration error of 0.724 cm in translation and 0.055 $^{\circ }$ in rotation in a single frame analysis with miscalibration magnitudes of up to $\pm$ 1.5 m and $\pm 20^{\circ }$ on the KITTI odometry dataset, which demonstrates the superiority of our method.
- 20 February 2023
- Published by Springer Nature in Neural Processing Letters
- Vol. 55 (5) , 6361-6375
Transformer-Based Cross-Modal Information Fusion Network for Semantic Segmentation
- 20 February 2023
- Published by Springer Nature in Neural Processing Letters
- Vol. 55 (5) , 6361-6375
Transformer-Based Cross-Modal Information Fusion Network for Semantic Segmentation
- Open Access
- 15 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 14 (19391404) , 4441-4451
Multimodal Urban Remote Sensing Image Registration Via Roadcross Triangular Feature
Automatic image registration of multimodal urban remote sensing images remains a critical challenging task in remote sensing image analysis due to significant nonlinear radiation distortions between multimodal image pairs; most of the traditional methods focus on the feature point detection and its local description and ignore the robust road information in multimodal urban remote sensing images. Motivated by this, we propose a fast and robust registration method for multimodal urban remote sensing images via road intersection triangular features. The proposed method obtains three main stages: Road lines extraction from images, intersection triangular feature construction, and triangular feature matching. The qualitative and quantitative experimental results show that the proposed method significantly outperforms other state-of-the-art methods, even when others completely fail to achieve the registration task of cross-modal images, our method still maintains good robustness and matching efficiency.
- 15 April 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 14 (19391404) , 4441-4451
Open AccessMultimodal Urban Remote Sensing Image Registration Via Roadcross Triangular Feature
Automatic image registration of multimodal urban remote sensing images remains a critical challenging task in remote sensing image analysis due to significant nonlinear radiation distortions between multimodal image pairs; most of the traditional methods focus on the feature point detection and its local description and ignore the robust road information in multimodal urban remote sensing images. Motivated by this, we propose a fast and robust registration method for multimodal urban remote sensing images via road intersection triangular features. The proposed method obtains three main stages: Road lines extraction from images, intersection triangular feature construction, and triangular feature matching. The qualitative and quantitative experimental results show that the proposed method significantly outperforms other state-of-the-art methods, even when others completely fail to achieve the registration task of cross-modal images, our method still maintains good robustness and matching efficiency.