Xiang Cao
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
Research Subjects
Xiang Cao has a total of 57 co-authors. Here are the top co-authors by number of shared publications:
- Changyin Sun
Southeast University, Nanjing, China
8 shared publications - Lu Ren
School of Artificial Intelligence and the Engineering Research Center of Autonomous Unmanned System Technology, Ministry of Education, Anhui University, Hefei, China
4 shared publications Changyin Sun
3 shared publicationsHongbing Sun
Huaiyin Normal Univ, Sch Phys & Elect Elect Engn, Huaian 223300, Peoples R China
3 shared publications- Xuerao Wang
Anhui University
3 shared publications Haoxiang Li
2 shared publications
Latest Publications
- 17 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 13 (3) , 3459-3475
Interpretable Hybrid Deep Reinforcement Learning-Based Energy Management in Low-Carbon Community Energy Systems With Temporal Attention Mechanism
This article proposes an interpretable deep reinforcement learning (DRL) method for energy management of low-carbon community energy systems (LCCES), which effectively addresses the transparency limitations caused by the black-box nature of traditional DRL neural network structures, thereby overcoming a key constraint in energy system applications. First, we develop a hybrid integer dynamic decision DRL algorithm to solve the low-carbon scheduling problem in community energy systems with continuous-discrete hybrid action spaces. Second, we construct an interpretable artificial intelligence framework, where the temporal attention mechanism is used to process and extract features to provide macro-level decision contribution analysis. These features are input into the decision tree for extracting device-level rules. Building upon this, we design an ensemble decision tree architecture with temporal attention mechanism to effectively identify critical time periods influenced by system inertia and energy fluctuations, thereby achieving interpretable optimization strategies while enhancing decision robustness under state fluctuations. Simulation results based on the independent test set demonstrate that, in comparison with alternative methods, the proposed approach yields a 22.1% cost reduction and a 32.4% carbon emission reduction rate relative to twin delayed deep deterministic policy gradient (TD3), and a 14.7% improvement in explanation accuracy compared with static decision trees.
- 17 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. 13 (3) , 3459-3475
Interpretable Hybrid Deep Reinforcement Learning-Based Energy Management in Low-Carbon Community Energy Systems With Temporal Attention Mechanism
This article proposes an interpretable deep reinforcement learning (DRL) method for energy management of low-carbon community energy systems (LCCES), which effectively addresses the transparency limitations caused by the black-box nature of traditional DRL neural network structures, thereby overcoming a key constraint in energy system applications. First, we develop a hybrid integer dynamic decision DRL algorithm to solve the low-carbon scheduling problem in community energy systems with continuous-discrete hybrid action spaces. Second, we construct an interpretable artificial intelligence framework, where the temporal attention mechanism is used to process and extract features to provide macro-level decision contribution analysis. These features are input into the decision tree for extracting device-level rules. Building upon this, we design an ensemble decision tree architecture with temporal attention mechanism to effectively identify critical time periods influenced by system inertia and energy fluctuations, thereby achieving interpretable optimization strategies while enhancing decision robustness under state fluctuations. Simulation results based on the independent test set demonstrate that, in comparison with alternative methods, the proposed approach yields a 22.1% cost reduction and a 32.4% carbon emission reduction rate relative to twin delayed deep deterministic policy gradient (TD3), and a 14.7% improvement in explanation accuracy compared with static decision trees.
- 7 February 2026
- Published by Springer Nature in Journal of Marine Science and Technology
- Vol. 31 (2) , 151-164
Encirclement advantage-guided improved multi-agent deep deterministic policy gradient for multi-target underwater pursuit–evasion
- 7 February 2026
- Published by Springer Nature in Journal of Marine Science and Technology
- Vol. 31 (2) , 151-164
Encirclement advantage-guided improved multi-agent deep deterministic policy gradient for multi-target underwater pursuit–evasion
- 6 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
OSP-SLAM: Semantic Prior-Free Visual SLAM for Arbitrary Dynamic Object
Visual Simultaneous Localization and Mapping (VSLAM) serves as a critical technology for mobile robots to achieve high-precision navigation and complex task execution. Traditional visual SLAM systems typically assume static environments, yet dynamic elements in dynamic scenarios interfere with data association, significantly degrading localization accuracy. To address this issue, this paper proposes OSP-SLAM, an improved system based on ORB-SLAM2. The system achieves dynamic region detection without prior knowledge through a fusion strategy integrating optical flow estimation, saliency region detection, and image segmentation. A lightweight CNN network is employed for feature point extraction and matching, combined with an adaptive Random Sample Consensus (RANSAC) algorithm to filter outliers, thereby maintaining feature robustness under degraded environments such as low illumination and excessively large dynamic regions. Experiments on the TUM RGB-D, Bonn RGB-D dynamic datasets, and real-world scenarios demonstrate that OSP-SLAM reduces the absolute trajectory error (ATE) by up to 98% compared to ORB-SLAM2. In complex dynamic scenarios, OSP-SLAM demonstrates significantly higher localization accuracy and robustness compared to state-of-the-art (SOTA) methods. It achieves 19% and 25% higher accuracy than RSO-SLAM and HMC-SLAM, respectively.
- 6 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
OSP-SLAM: Semantic Prior-Free Visual SLAM for Arbitrary Dynamic Object
Visual Simultaneous Localization and Mapping (VSLAM) serves as a critical technology for mobile robots to achieve high-precision navigation and complex task execution. Traditional visual SLAM systems typically assume static environments, yet dynamic elements in dynamic scenarios interfere with data association, significantly degrading localization accuracy. To address this issue, this paper proposes OSP-SLAM, an improved system based on ORB-SLAM2. The system achieves dynamic region detection without prior knowledge through a fusion strategy integrating optical flow estimation, saliency region detection, and image segmentation. A lightweight CNN network is employed for feature point extraction and matching, combined with an adaptive Random Sample Consensus (RANSAC) algorithm to filter outliers, thereby maintaining feature robustness under degraded environments such as low illumination and excessively large dynamic regions. Experiments on the TUM RGB-D, Bonn RGB-D dynamic datasets, and real-world scenarios demonstrate that OSP-SLAM reduces the absolute trajectory error (ATE) by up to 98% compared to ORB-SLAM2. In complex dynamic scenarios, OSP-SLAM demonstrates significantly higher localization accuracy and robustness compared to state-of-the-art (SOTA) methods. It achieves 19% and 25% higher accuracy than RSO-SLAM and HMC-SLAM, respectively.
- 28 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 26 (7) , 9461-9474
Collision Avoidance Based on Stochastic Model Predictive Control in Collaboration Between ROV and AUV
Due to certain technical limitations of autonomous underwater vehicle (AUV), they cannot completely perform complex tasks independently. When performing complex tasks, coordination between the remote operated vehicle (ROV) and AUV is required. Therefore, collision avoidance is a key technology to ensure vehicle safety. During collision avoidance, AUV need to understand human intentions, make decisions, and perform the corresponding actions. To solve the problems of human intention uncertainty and random noise interference, an AUV collision avoidance strategy based on a dynamic Bayesian network and stochastic model predictive control (SMPC) is proposed in this paper. First, a dynamic Bayesian network is used to assess the probability of AUV collisions in the system. Then, using the properties of Gaussian distribution and related theorems, the objective function is simplified and transformed into a deterministic model predictive control problem. Finally, the intention-exploration item is added to the objective function to better understand human intention. Through the simulations and experiments in specific scenarios, it is verified that the proposed collision avoidance control strategy can safely and effectively control a hybrid system with the coexistence of ROV and AUV.
- 28 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 26 (7) , 9461-9474
Collision Avoidance Based on Stochastic Model Predictive Control in Collaboration Between ROV and AUV
Due to certain technical limitations of autonomous underwater vehicle (AUV), they cannot completely perform complex tasks independently. When performing complex tasks, coordination between the remote operated vehicle (ROV) and AUV is required. Therefore, collision avoidance is a key technology to ensure vehicle safety. During collision avoidance, AUV need to understand human intentions, make decisions, and perform the corresponding actions. To solve the problems of human intention uncertainty and random noise interference, an AUV collision avoidance strategy based on a dynamic Bayesian network and stochastic model predictive control (SMPC) is proposed in this paper. First, a dynamic Bayesian network is used to assess the probability of AUV collisions in the system. Then, using the properties of Gaussian distribution and related theorems, the objective function is simplified and transformed into a deterministic model predictive control problem. Finally, the intention-exploration item is added to the objective function to better understand human intention. Through the simulations and experiments in specific scenarios, it is verified that the proposed collision avoidance control strategy can safely and effectively control a hybrid system with the coexistence of ROV and AUV.
- 19 March 2025
- Published by Cambridge University Press (CUP) in Robotica
- Vol. 43 (4) , 1331-1349
Mobile robot tracking control based on lightweight network
Target tracking technology is a key research area in the field of mobile robots, with wide applications in logistics, security, autonomous driving, and more. It generally involves two main components: target recognition and target following. However, the limited computational power of the mobile robot’s controller makes achieving high precision and fast target recognition and tracking a challenge. To address the challenges posed by limited computing power, this paper proposes a target-tracking control algorithm based on lightweight neural networks. First, a depthwise separable convolution-based backbone is introduced for feature extraction. Then, an efficient channel attention module is incorporated into the target recognition algorithm to minimize the impact of redundant features and emphasize important channels, thereby reducing model complexity and enhancing network efficiency. Finally, based on the data collected from visual and ultrasonic sensors, a model predictive control strategy is used to achieve target tracking. Validation of the proposed algorithm is conducted using a mobile robot equipped with Raspberry Pi 4B. Experimental results demonstrate that the proposed algorithm achieves rapid target tracking.
- 19 March 2025
- Published by Cambridge University Press (CUP) in Robotica
- Vol. 43 (4) , 1331-1349
Mobile robot tracking control based on lightweight network
Target tracking technology is a key research area in the field of mobile robots, with wide applications in logistics, security, autonomous driving, and more. It generally involves two main components: target recognition and target following. However, the limited computational power of the mobile robot’s controller makes achieving high precision and fast target recognition and tracking a challenge. To address the challenges posed by limited computing power, this paper proposes a target-tracking control algorithm based on lightweight neural networks. First, a depthwise separable convolution-based backbone is introduced for feature extraction. Then, an efficient channel attention module is incorporated into the target recognition algorithm to minimize the impact of redundant features and emphasize important channels, thereby reducing model complexity and enhancing network efficiency. Finally, based on the data collected from visual and ultrasonic sensors, a model predictive control strategy is used to achieve target tracking. Validation of the proposed algorithm is conducted using a mobile robot equipped with Raspberry Pi 4B. Experimental results demonstrate that the proposed algorithm achieves rapid target tracking.
- 23 November 2024
- Published by Elsevier in Ocean Engineering
- Vol. 315
Path Re-planning method of unmanned underwater vehicles based on dynamic bayesian threat assessment
- 23 November 2024
- Published by Elsevier in Ocean Engineering
- Vol. 315
Path Re-planning method of unmanned underwater vehicles based on dynamic bayesian threat assessment
- 17 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Vehicles
- Vol. 10 (3) , 1568-1580
AUV Collision Avoidance Strategy Based on Fuzzy Reinforcement Learning
Collision avoidance is a key technology for autonomous underwater vehicle (AUV) to achieve tasks such as path planning, target searching, and map construction. The performance of the algorithm directly affects the safety of the AUV and the success of collision avoidance. To improve the algorithm's generalization and adaptability, a fuzzy reinforcement learning collision avoidance strategy is presented in this study. Firstly, reinforcement learning learns without explicit modeling and converge quickly, making it suitable for continuous online learning tasks. Therefore, this paper adopts a multi-step temporal difference reinforcement learning approach to control the collision avoidance of AUV. Then, fuzzy theory is integrated into reinforcement learning to address generalization issues in the state space. This integration allows the acquisition of continuous state inputs and action outputs, enhancing motion state recognition and improving operational smoothness. Finally, enhancing online training of the fuzzy reinforcement learning method through adaptive strategy adjustments has improved the algorithm's online optimization efficiency. The proposed AUV collision avoidance strategy is validated through simulations and experiments in a three-dimensional underwater environment. The research results demonstrate that this strategy can safety and efficiently guide the AUV, showcasing strong generalization capabilities.
- 17 July 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Vehicles
- Vol. 10 (3) , 1568-1580
AUV Collision Avoidance Strategy Based on Fuzzy Reinforcement Learning
Collision avoidance is a key technology for autonomous underwater vehicle (AUV) to achieve tasks such as path planning, target searching, and map construction. The performance of the algorithm directly affects the safety of the AUV and the success of collision avoidance. To improve the algorithm's generalization and adaptability, a fuzzy reinforcement learning collision avoidance strategy is presented in this study. Firstly, reinforcement learning learns without explicit modeling and converge quickly, making it suitable for continuous online learning tasks. Therefore, this paper adopts a multi-step temporal difference reinforcement learning approach to control the collision avoidance of AUV. Then, fuzzy theory is integrated into reinforcement learning to address generalization issues in the state space. This integration allows the acquisition of continuous state inputs and action outputs, enhancing motion state recognition and improving operational smoothness. Finally, enhancing online training of the fuzzy reinforcement learning method through adaptive strategy adjustments has improved the algorithm's online optimization efficiency. The proposed AUV collision avoidance strategy is validated through simulations and experiments in a three-dimensional underwater environment. The research results demonstrate that this strategy can safety and efficiently guide the AUV, showcasing strong generalization capabilities.
- 24 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Vehicles
- Vol. 10 (2) , 739-751
State Recovery and Fault-Tolerant Control of Autonomous Surface Vehicle With Actuator and Sensor Faults
This paper is concerned with the trajectory tracking problem for autonomous surface vehicles (ASVs) subject to simultaneous actuator and sensor faults. An output-feedback multiple fault-tolerant control (MFTC) scheme is proposed by using inaccurate position-heading measurement only. First, a modified extended state observer (ESO) is proposed to simultaneously achieve the state recovery of position-heading and non-measurable velocities under unknown dynamics and external disturbances. To eliminate the effect of multiple faults, separated adaptive compensation mechanisms are introduced to estimate various fault factors individually. Based on the proposed ESO and estimated factors, an adaptive output-feedback MFTC strategy is explored to stabilize the closed-loop control system in the presence of multiple faults. It is proven that the proposed controller guarantees that all signals in the closed-loop control system are bounded, including tracking error, observer error, and estimation errors. Finally, simulation results with several cases are provided to demonstrate the effectiveness of the proposed MFTC method.
- 24 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Vehicles
- Vol. 10 (2) , 739-751
State Recovery and Fault-Tolerant Control of Autonomous Surface Vehicle With Actuator and Sensor Faults
This paper is concerned with the trajectory tracking problem for autonomous surface vehicles (ASVs) subject to simultaneous actuator and sensor faults. An output-feedback multiple fault-tolerant control (MFTC) scheme is proposed by using inaccurate position-heading measurement only. First, a modified extended state observer (ESO) is proposed to simultaneously achieve the state recovery of position-heading and non-measurable velocities under unknown dynamics and external disturbances. To eliminate the effect of multiple faults, separated adaptive compensation mechanisms are introduced to estimate various fault factors individually. Based on the proposed ESO and estimated factors, an adaptive output-feedback MFTC strategy is explored to stabilize the closed-loop control system in the presence of multiple faults. It is proven that the proposed controller guarantees that all signals in the closed-loop control system are bounded, including tracking error, observer error, and estimation errors. Finally, simulation results with several cases are provided to demonstrate the effectiveness of the proposed MFTC method.