Heqing Li
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
Heqing Li has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Xinde Li
School of Automation, Southeast University, Nanjing, China
5 shared publications - Fir Dunkin
Key Laboratory of Measurement and Control of CSE, School of Automation, Southeast University, Nanjing, China
4 shared publications - Chuanfei Hu
Faculty of Intelligence Technology, Shanghai Institute of Technology, Shanghai, 201418, China
3 shared publications - Shuzhi Sam Ge
Electrical and Computer Engineering, National University of Singapore, Singapore, Singapore
3 shared publications - Zhentong Zhang
School of Cyber Science and Engineering, Southeast University, Nanjing, China
3 shared publications - Guoliang Wu
Key Laboratory of Measurement and Control of CSE, School of Automation, Southeast University, Nanjing, China
2 shared publications
Latest Publications
- 1 March 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 339
Virus-RK: A bionic adversarial attack framework for visual data in UAV 3D reconstruction system
- 1 March 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 339
Virus-RK: A bionic adversarial attack framework for visual data in UAV 3D reconstruction system
- 7 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 7748-7761
MgCNL: A Sample Separation Approach via Multi-Granularity Balls for Fault Diagnosis With the Interference of Noisy Labels
The fault diagnosis based on supervised learning has achieved remarkable results in the intelligent manufacturing, making it an important guarantee for long-term safe and stable operation in modern industry. However, the accuracy heavily relies on high-quality annotation labels, which are expensive to obtain, limiting the diagnosis models applicability in many scenarios. Although obtaining automatically annotated samples from annotators is a promising solution, the generated dataset is always containing incorrect labels (noisy labels), due to perceptual limitations, resulting in low or even invalid the accuracy of model. With the goal of handling this challenge, a diagnostic approach based on multi-granularity information fusion to combat noisy labels, called MgCNL, is proposed, to train the model with high-accuracy, without knowing the specific noise ratio. Specifically, inspired by granular-ball computing, a confidence evaluation method of labels is designed, so that samples with high confidence labels can be selected from dataset with noisy labels for supervised learning, thus avoiding the negative impact of incorrect labels on model performance. Finally, the efficacy was demonstrated on three datasets using different backbones: MgCNL successfully reduced the adverse impact of noisy labels, achieving significantly better results than other advanced methods in various noisy scenarios, which offers a competitive model training strategy for practitioners in intelligent manufacturing or industrial fault diagnosis who are hampered by the costs associated with sample labeling. Note to Practitioners —In modern industry, the cost of manual/expert annotation for high-quality data is is prohibitively expensive, and the data annotated by automatic annotators often contains noisy labels that seriously damages the accuracy of models, which makes many data-driven diagnosis models constrained by training data and difficult to put into practice, posing an urgent challenge to the automation and intelligence of the manufacturing industry. To address this challenge, this article proposed a robust training strategy called MgCNL, aimed at offsetting the negative impact of noisy labels, in the hope that automatic annotation strategy with lower cost can be more widely applied in model training tasks for industrial practice. MgCNL, based on multi-granularity information, can effectively select high-confidence samples from datasets for supervised learning, even under unknown proportions of noise labels, thus reducing the misleading impact of noisy labels on diagnostic models. As a result, MgCNL possesses the ability to robustly train high-accuracy diagnostic models in data with noisy labels, thus enabling automatic annotators to replace experts in dataset construction as a more economical and efficient potential technical approach. Meanwhile, MgCNL also brings value to datasets with uncertain labels, making them applicable without the need to invest significant human resources to verify label reliability.
- 7 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 7748-7761
MgCNL: A Sample Separation Approach via Multi-Granularity Balls for Fault Diagnosis With the Interference of Noisy Labels
The fault diagnosis based on supervised learning has achieved remarkable results in the intelligent manufacturing, making it an important guarantee for long-term safe and stable operation in modern industry. However, the accuracy heavily relies on high-quality annotation labels, which are expensive to obtain, limiting the diagnosis models applicability in many scenarios. Although obtaining automatically annotated samples from annotators is a promising solution, the generated dataset is always containing incorrect labels (noisy labels), due to perceptual limitations, resulting in low or even invalid the accuracy of model. With the goal of handling this challenge, a diagnostic approach based on multi-granularity information fusion to combat noisy labels, called MgCNL, is proposed, to train the model with high-accuracy, without knowing the specific noise ratio. Specifically, inspired by granular-ball computing, a confidence evaluation method of labels is designed, so that samples with high confidence labels can be selected from dataset with noisy labels for supervised learning, thus avoiding the negative impact of incorrect labels on model performance. Finally, the efficacy was demonstrated on three datasets using different backbones: MgCNL successfully reduced the adverse impact of noisy labels, achieving significantly better results than other advanced methods in various noisy scenarios, which offers a competitive model training strategy for practitioners in intelligent manufacturing or industrial fault diagnosis who are hampered by the costs associated with sample labeling. Note to Practitioners —In modern industry, the cost of manual/expert annotation for high-quality data is is prohibitively expensive, and the data annotated by automatic annotators often contains noisy labels that seriously damages the accuracy of models, which makes many data-driven diagnosis models constrained by training data and difficult to put into practice, posing an urgent challenge to the automation and intelligence of the manufacturing industry. To address this challenge, this article proposed a robust training strategy called MgCNL, aimed at offsetting the negative impact of noisy labels, in the hope that automatic annotation strategy with lower cost can be more widely applied in model training tasks for industrial practice. MgCNL, based on multi-granularity information, can effectively select high-confidence samples from datasets for supervised learning, even under unknown proportions of noise labels, thus reducing the misleading impact of noisy labels on diagnostic models. As a result, MgCNL possesses the ability to robustly train high-accuracy diagnostic models in data with noisy labels, thus enabling automatic annotators to replace experts in dataset construction as a more economical and efficient potential technical approach. Meanwhile, MgCNL also brings value to datasets with uncertain labels, making them applicable without the need to invest significant human resources to verify label reliability.
- 18 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-13
Trust Measurement of Visual Data Based on Multigranularity Belief Fusion for UAV Perception System
Unmanned aerial vehicle (UAV) visual perception plays a pivotal role in the realm of intelligent big data applications such as smart transportation, urban planning, and precision agriculture. Ensuring the integrity and reliability of data collected by UAVs is essential to prevent potential misinformation and tampering in critical data acquisition tasks. However, conventional single assessment and decision-making mechanisms have demonstrated limitations, leaving systems vulnerable to security threats in realworld scenarios. In this study, we propose an innovative trust measurement framework based on multi-granularity belief fusion (TM-MGBF), aimed at bolstering the reliability of UAV visual data and mitigating the risks posed by malicious attacks. The integration of multi-granularity trust measurement offers enhanced adaptability and accuracy, offering a promising avenue to address the aforementioned challenges. Specifically, our approach involves devising distinct methodologies for assessing trust across diverse data sources to ascertain their trustworthiness. Furthermore, we introduce an improved fusion strategy, achieving more rational trust measurement by weighting the importance assessment of trust evidence. Experimental results validate the efficacy of the proposed trust model based on multi-granularity belief fusion, showcasing its superior performance in the evaluation of visual perception data and the fortification of UAV perception systems’ security posture.
- 18 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-13
Trust Measurement of Visual Data Based on Multigranularity Belief Fusion for UAV Perception System
Unmanned aerial vehicle (UAV) visual perception plays a pivotal role in the realm of intelligent big data applications such as smart transportation, urban planning, and precision agriculture. Ensuring the integrity and reliability of data collected by UAVs is essential to prevent potential misinformation and tampering in critical data acquisition tasks. However, conventional single assessment and decision-making mechanisms have demonstrated limitations, leaving systems vulnerable to security threats in realworld scenarios. In this study, we propose an innovative trust measurement framework based on multi-granularity belief fusion (TM-MGBF), aimed at bolstering the reliability of UAV visual data and mitigating the risks posed by malicious attacks. The integration of multi-granularity trust measurement offers enhanced adaptability and accuracy, offering a promising avenue to address the aforementioned challenges. Specifically, our approach involves devising distinct methodologies for assessing trust across diverse data sources to ascertain their trustworthiness. Furthermore, we introduce an improved fusion strategy, achieving more rational trust measurement by weighting the importance assessment of trust evidence. Experimental results validate the efficacy of the proposed trust model based on multi-granularity belief fusion, showcasing its superior performance in the evaluation of visual perception data and the fortification of UAV perception systems’ security posture.
- 12 September 2024
- Published by Elsevier in Computers & Security
- Vol. 148
Adaptive multi-granularity trust management scheme for UAV visual sensor security under adversarial attacks
- 12 September 2024
- Published by Elsevier in Computers & Security
- Vol. 148
Adaptive multi-granularity trust management scheme for UAV visual sensor security under adversarial attacks
- 1 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 62 (01962892) , 1-13
Dual-Branch Sparse Self-Learning With Instance Binding Augmentation for Adversarial Detection in Remote Sensing Images
Remote sensing image analysis technology based on neural networks has significantly facilitated human life. However, adversarial attacks can drastically impair the performance of these models, posing substantial economic and security risks. Current adversarial example (AE) detectors primarily focus on studying attacked natural images, while AEs in the remote sensing domain have not received adequate attention. To address this challenge, we propose a novel dual-branch sparse self-learning framework, leveraging instance binding augmentation. The contrastive branch concurrently enhances intra-instance and inter-example feature discrimination, while the masked branch reconstructs perturbation distributions. Furthermore, our method utilizes sparse encoding within depthwise separable convolutions to efficiently transfer parameters, thereby ensuring compatibility with deployment on mobile devices. Extensive experiments demonstrate that our method achieves state-of-the-art performance in detecting both white-box and black-box attacks on remote sensing images. Specifically, our method achieves an average detection accuracy of 95.69%/95.18%, a recall of 91.8%/93.6%, and an F1 score of 93.48%/94.22% on two attacked models across various attack scenarios, outperforming existing methods.
- 1 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 62 (01962892) , 1-13
Dual-Branch Sparse Self-Learning With Instance Binding Augmentation for Adversarial Detection in Remote Sensing Images
Remote sensing image analysis technology based on neural networks has significantly facilitated human life. However, adversarial attacks can drastically impair the performance of these models, posing substantial economic and security risks. Current adversarial example (AE) detectors primarily focus on studying attacked natural images, while AEs in the remote sensing domain have not received adequate attention. To address this challenge, we propose a novel dual-branch sparse self-learning framework, leveraging instance binding augmentation. The contrastive branch concurrently enhances intra-instance and inter-example feature discrimination, while the masked branch reconstructs perturbation distributions. Furthermore, our method utilizes sparse encoding within depthwise separable convolutions to efficiently transfer parameters, thereby ensuring compatibility with deployment on mobile devices. Extensive experiments demonstrate that our method achieves state-of-the-art performance in detecting both white-box and black-box attacks on remote sensing images. Specifically, our method achieves an average detection accuracy of 95.69%/95.18%, a recall of 91.8%/93.6%, and an F1 score of 93.48%/94.22% on two attacked models across various attack scenarios, outperforming existing methods.
- Open Access
- 18 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 76-79
A Path Planning Framework for Air-land Amphibious UAV with Fast Response and Fine Learning
Air-land amphibious unmanned aerial vehicle (ALAUAV), which integrate ground mobility and aerial flight capabilities, are versatile intelligent entities. Efficient global and precise local path planning is of paramount significance in ensuring the high-quality execution of their missions, as well as system safety and stability. However, current methodologies often focus on single aspects, resulting in complex path computations and inadequate precision. In this study, we propose a path planning framework based on the integration of multiple perspectives and reinforcement learning. In addressing known environments, we transform intricate three-dimensional (3D) terrains into two-dimensional (2D) multi-view representations and perform parallel computations, effectively resolving issues related to delayed responsiveness and reduced efficiency. Subsequently, for dynamic and unknown environments, we establish a deep reinforcement learning model to achieve local fine-grained path planning. To demonstrate the effectiveness of our approach, we conduct simulation experiments in both known and unknown environments. The experimental results unequivocally demonstrate the superior performance of our path planning method in the context of ALAUAV, laying a robust foundation for further expanding the application of this technology in complex scenarios.
- 18 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 76-79
Open AccessA Path Planning Framework for Air-land Amphibious UAV with Fast Response and Fine Learning
Air-land amphibious unmanned aerial vehicle (ALAUAV), which integrate ground mobility and aerial flight capabilities, are versatile intelligent entities. Efficient global and precise local path planning is of paramount significance in ensuring the high-quality execution of their missions, as well as system safety and stability. However, current methodologies often focus on single aspects, resulting in complex path computations and inadequate precision. In this study, we propose a path planning framework based on the integration of multiple perspectives and reinforcement learning. In addressing known environments, we transform intricate three-dimensional (3D) terrains into two-dimensional (2D) multi-view representations and perform parallel computations, effectively resolving issues related to delayed responsiveness and reduced efficiency. Subsequently, for dynamic and unknown environments, we establish a deep reinforcement learning model to achieve local fine-grained path planning. To demonstrate the effectiveness of our approach, we conduct simulation experiments in both known and unknown environments. The experimental results unequivocally demonstrate the superior performance of our path planning method in the context of ALAUAV, laying a robust foundation for further expanding the application of this technology in complex scenarios.
- Open Access
- 12 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 119-125
Analysis of Airframe Parameters of OAT-bicopter to Optimize Pitch Stability
Unmanned aerial vehicle (UAV), a type of airborne mobile robot operated without human intervention, has become a research hotspot in recent years due to its numerous advantages. Among them, bicopter has attracted much attention for its superior maneuverability. However, the flight performance of the bicopter is easily affected by various airframe parameters, such as the position of the center of gravity, rotor inertia, and angle of the servo tilting axis. In this study, we conduct theoretical analysis of the Oblique Active Tilting(OAT) bicopter to identify optimal parameters for stable flight. First, we obtain the transfer function of the UAV’s pitch control system based on its pitch motion equation, and analyze its pitch stability using the characteristic equation. Then, we investigate the influence of parameters such as the position of the center of gravity, rotor inertia, and angle of the servo tilting axis on the UAV’s dynamic and steady-state performance, and propose a method to determine the optimal angle of the servo tilting axis. Finally, we validate the effectiveness of our proposed parameters through simulation experiments, demonstrating the theoretical and practical significance of our research for stable control and application of OAT-bicopters.
- 12 May 2023
- Published by Association for Computing Machinery (ACM)
- p. 119-125
Open AccessAnalysis of Airframe Parameters of OAT-bicopter to Optimize Pitch Stability
Unmanned aerial vehicle (UAV), a type of airborne mobile robot operated without human intervention, has become a research hotspot in recent years due to its numerous advantages. Among them, bicopter has attracted much attention for its superior maneuverability. However, the flight performance of the bicopter is easily affected by various airframe parameters, such as the position of the center of gravity, rotor inertia, and angle of the servo tilting axis. In this study, we conduct theoretical analysis of the Oblique Active Tilting(OAT) bicopter to identify optimal parameters for stable flight. First, we obtain the transfer function of the UAV’s pitch control system based on its pitch motion equation, and analyze its pitch stability using the characteristic equation. Then, we investigate the influence of parameters such as the position of the center of gravity, rotor inertia, and angle of the servo tilting axis on the UAV’s dynamic and steady-state performance, and propose a method to determine the optimal angle of the servo tilting axis. Finally, we validate the effectiveness of our proposed parameters through simulation experiments, demonstrating the theoretical and practical significance of our research for stable control and application of OAT-bicopters.
- 28 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 20 (1) , 732-743
ESUAV-NI: Endogenous Security Framework for UAV Perception System Based on Neural Immunity
Unmanned aerial vehicles (UAVs) represent an essential component of advanced intelligent equipment that can be used as an aerial perception system by installing various sensors such as vision, hearing, touch, taste, and smell to achieve intelligently integrated perception of environments. However, these perception system with environmental information may be threatened by various internal and external attacks, causing a great challenge to the security of the UAV. The original security system relied on an expert knowledge base to prevent attacks, but the weaknesses of lacking proactivity and flexibility are gradually exposed. The strong resistance and survivability of biological systems can be used to fill this capability gap and provide new ideas for the security of the UAV perception system. Therefore, an endogenous security framework (ESUAV-NI) based on the neural system and immune system is proposed in this article. Through breeding artificial intelligence (AI) vaccines and distributed neural hierarchical control, we achieve the security protection for the UAV perception system. Moreover, we evaluated the AI vaccine breeding approach in the ESUAV-NI by conducting extensive experiments on internal threats and external aerial imagery camouflage data, respectively. The results show that the proposed approach has a superior performance for the UAV perception system.
- 28 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 20 (1) , 732-743
ESUAV-NI: Endogenous Security Framework for UAV Perception System Based on Neural Immunity
Unmanned aerial vehicles (UAVs) represent an essential component of advanced intelligent equipment that can be used as an aerial perception system by installing various sensors such as vision, hearing, touch, taste, and smell to achieve intelligently integrated perception of environments. However, these perception system with environmental information may be threatened by various internal and external attacks, causing a great challenge to the security of the UAV. The original security system relied on an expert knowledge base to prevent attacks, but the weaknesses of lacking proactivity and flexibility are gradually exposed. The strong resistance and survivability of biological systems can be used to fill this capability gap and provide new ideas for the security of the UAV perception system. Therefore, an endogenous security framework (ESUAV-NI) based on the neural system and immune system is proposed in this article. Through breeding artificial intelligence (AI) vaccines and distributed neural hierarchical control, we achieve the security protection for the UAV perception system. Moreover, we evaluated the AI vaccine breeding approach in the ESUAV-NI by conducting extensive experiments on internal threats and external aerial imagery camouflage data, respectively. The results show that the proposed approach has a superior performance for the UAV perception system.
- Open Access
- 13 December 2019
- Published by Springer Nature in Scientific Reports
- Vol. 9 (1) , 1-10
Optically modulated magnetic resonance of erbium implanted silicon
Er implanted Si is a candidate for quantum and photonic applications; however, several different Er centres are generated, and their symmetry, energy level structure, magnetic and optical properties, and mutual interactions have been poorly understood, which has been a major barrier to the development of these applications. Optically modulated magnetic resonance (OMMR) gives a spectrum of the modulation of an electron paramagnetic resonance (EPR) signal by a tuneable optical field. Our OMMR spectrum of Er implanted Si agrees with three independent measurements, showing that we have made the first measurement of the crystal field splitting of the 4I13/2 manifold of Er implanted Si, and allows us to revise the crystal field splitting of the 4I15/2 manifold. This splitting originates from a photoluminescence (PL) active O coordinated Er centre with orthorhombic C2v symmetry, which neighbours an EPR active O coordinated Er centre with monoclinic C1h symmetry. This pair of centres could form the basis of a controlled NOT (CNOT) gate.
- 13 December 2019
- Published by Springer Nature in Scientific Reports
- Vol. 9 (1) , 1-10
Open AccessOptically modulated magnetic resonance of erbium implanted silicon
Er implanted Si is a candidate for quantum and photonic applications; however, several different Er centres are generated, and their symmetry, energy level structure, magnetic and optical properties, and mutual interactions have been poorly understood, which has been a major barrier to the development of these applications. Optically modulated magnetic resonance (OMMR) gives a spectrum of the modulation of an electron paramagnetic resonance (EPR) signal by a tuneable optical field. Our OMMR spectrum of Er implanted Si agrees with three independent measurements, showing that we have made the first measurement of the crystal field splitting of the 4I13/2 manifold of Er implanted Si, and allows us to revise the crystal field splitting of the 4I15/2 manifold. This splitting originates from a photoluminescence (PL) active O coordinated Er centre with orthorhombic C2v symmetry, which neighbours an EPR active O coordinated Er centre with monoclinic C1h symmetry. This pair of centres could form the basis of a controlled NOT (CNOT) gate.