Weili Ding
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Weili Ding has a total of 118 co-authors. Here are the top co-authors by number of shared publications:
- Changchun Hua
Yanshan University
18 shared publications - Ying Zhang
Yanshan University
7 shared publications - Cui-Hua Zhang
Yanshan University
5 shared publications - Guanglei Zhao
Institute of Electrical Engineering, Key Lab of Intelligent Rehabilitation and Neuroregulation in Hebei Province, and Engineering Research Center of Intelligent Control System and Intelligent Equipment, Ministry of Education, Yanshan University, Qinhuangdao, China
5 shared publications Changchun Hua
Key Lab of Intelligent Rehabilitation and Neuroregulation in Hebei Province, Yanshan University, Qinhuangdao 066004, People’s Republic of China
4 shared publications- Biao Lu
Nankai University
3 shared publications
Latest Publications
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10 (5) , 3543-3554
SAT-GCN: Spatial-Aware Transformer-Enhanced Graph Convolutional Network for Skeleton-Based Action Recognition
In the domain of skeleton-based action recognition, Graph Convolutional Networks (GCNs) approaches have achieved remarkable progress, substantially advancing the field. Although recent GCN-based approaches have introduced dynamic topology modeling and the learning of multi-scale or long-range dependencies, the capture of distant joint interactions and non-physical semantic relationships remains fundamentally constrained by the inherent limitations of skeletal connectivity. As a consequence, the recognition accuracy of actions requiring global motion coordination or involving fine-grained movements is adversely affected. To address this challenge, we propose a novel Spatial-Aware Transformer Graph Convolution (SAT-GC) module. This module effectively integrates the global attention mechanism of Transformers with the local relational modeling capabilities of GCNs. Building upon this module, a powerful graph convolutional network, termed SAT-GCN, is further developed using a progressive modeling strategy. The proposed network retains the inherent local skeletal topology and simultaneously strengthens the ability to model dynamic and long-range dependencies among joints. This design enables more comprehensive spatial feature extraction from complex action sequences. The proposed SAT-GCN has been evaluated on four benchmark datasets: NTU RGB+D, NTU RGB+D 120, NW-UCLA, and Kinetics-Skeleton 400. Extensive experimental results have demonstrated that SAT-GCN consistently outperforms state-of-the-art methods across evaluation settings.
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 10 (5) , 3543-3554
SAT-GCN: Spatial-Aware Transformer-Enhanced Graph Convolutional Network for Skeleton-Based Action Recognition
In the domain of skeleton-based action recognition, Graph Convolutional Networks (GCNs) approaches have achieved remarkable progress, substantially advancing the field. Although recent GCN-based approaches have introduced dynamic topology modeling and the learning of multi-scale or long-range dependencies, the capture of distant joint interactions and non-physical semantic relationships remains fundamentally constrained by the inherent limitations of skeletal connectivity. As a consequence, the recognition accuracy of actions requiring global motion coordination or involving fine-grained movements is adversely affected. To address this challenge, we propose a novel Spatial-Aware Transformer Graph Convolution (SAT-GC) module. This module effectively integrates the global attention mechanism of Transformers with the local relational modeling capabilities of GCNs. Building upon this module, a powerful graph convolutional network, termed SAT-GCN, is further developed using a progressive modeling strategy. The proposed network retains the inherent local skeletal topology and simultaneously strengthens the ability to model dynamic and long-range dependencies among joints. This design enables more comprehensive spatial feature extraction from complex action sequences. The proposed SAT-GCN has been evaluated on four benchmark datasets: NTU RGB+D, NTU RGB+D 120, NW-UCLA, and Kinetics-Skeleton 400. Extensive experimental results have demonstrated that SAT-GCN consistently outperforms state-of-the-art methods across evaluation settings.
- 4 March 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (6)
A General Two-Stage Framework for Truck Bucket Pose Estimation via Monocular Vision
Accurately estimating the 6D pose of the truck bucket is a critical challenge in the field of remotely controlled construction machinery, with significant implications for improving operational safety and efficiency. To address perception deficiencies and low efficiency in remote manual operations, this study proposes a monocular camera-based two-stage 6D pose estimation method for the truck bucket. Unlike traditional 3D object detection methods that require labeled point cloud data for training, this approach only necessitates simple circular video capture around the truck bucket to construct a sparse Structure-from-Motion (SfM) model. Subsequently, an aggregation-attention layer is innovatively employed to fuse associated 2D features into 3D feature descriptors. By leveraging self-attention and cross-attention layers, the 2D features of the query image are matched with the 3D features of the point cloud, enabling real-time 6D pose estimation without task-specific training. Experiments conducted on real-world dump trucks demonstrate that the proposed method achieves both high accuracy and efficiency. Specifically, the method achieves a single-frame localization time of only 0.0297 seconds and a video processing speed of approximately 10 FPS for 1920×1080 resolution videos. Under the 20cm-20deg threshold, the pose estimation success rate reaches 93.8%. Furthermore, the method demonstrates excellent generalization capabilities in unseen complex engineering scenarios. These results indicate that the proposed method provides significant technical support for advancing the intelligence and efficiency of remotely controlled construction machinery operations.
- 4 March 2026
- Published by IOP Publishing in Engineering Research Express
- Vol. 8 (6)
A General Two-Stage Framework for Truck Bucket Pose Estimation via Monocular Vision
Accurately estimating the 6D pose of the truck bucket is a critical challenge in the field of remotely controlled construction machinery, with significant implications for improving operational safety and efficiency. To address perception deficiencies and low efficiency in remote manual operations, this study proposes a monocular camera-based two-stage 6D pose estimation method for the truck bucket. Unlike traditional 3D object detection methods that require labeled point cloud data for training, this approach only necessitates simple circular video capture around the truck bucket to construct a sparse Structure-from-Motion (SfM) model. Subsequently, an aggregation-attention layer is innovatively employed to fuse associated 2D features into 3D feature descriptors. By leveraging self-attention and cross-attention layers, the 2D features of the query image are matched with the 3D features of the point cloud, enabling real-time 6D pose estimation without task-specific training. Experiments conducted on real-world dump trucks demonstrate that the proposed method achieves both high accuracy and efficiency. Specifically, the method achieves a single-frame localization time of only 0.0297 seconds and a video processing speed of approximately 10 FPS for 1920×1080 resolution videos. Under the 20cm-20deg threshold, the pose estimation success rate reaches 93.8%. Furthermore, the method demonstrates excellent generalization capabilities in unseen complex engineering scenarios. These results indicate that the proposed method provides significant technical support for advancing the intelligence and efficiency of remotely controlled construction machinery operations.
- 3 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 22 (6) , 5095-5105
Adaptive PID Coupling Tracking Control for Double-Pendulum Crane Based on Digital Twin Technology
This article presents an advanced digital twin system constructed for a double-pendulum crane (DPC) with complex dynamic characteristics, aiming to address the pressing need for intelligent control of DPC. The high complexity and underactuated nature of the DPC system make virtual-physical mapping, tracking control, and antiswing control highly challenging. To tackle these issues, this study designs a novel adaptive proportional-integral-derivative (PID) coupling tracking control method with an automatic gain adjustment mechanism, achieving real-time and accurate motion mapping from the physical to the virtual DPC and effectively suppressing the double-pendulum swing, which is a key goal. Particularly, to effectively compensate for different lumped disturbances in both the physical and virtual DPC and improve tracking accuracy, a time-delay control strategy is integrated into the underactuated DPC system, and an adaptive method is employed to estimate the influence of state-dependent time-delay errors online, thereby relaxing conventional boundedness assumptions. Finally, synchronous motion experiments demonstrate the effectiveness and superiority of the proposed virtual-physical tracking control method.
- 3 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 22 (6) , 5095-5105
Adaptive PID Coupling Tracking Control for Double-Pendulum Crane Based on Digital Twin Technology
This article presents an advanced digital twin system constructed for a double-pendulum crane (DPC) with complex dynamic characteristics, aiming to address the pressing need for intelligent control of DPC. The high complexity and underactuated nature of the DPC system make virtual-physical mapping, tracking control, and antiswing control highly challenging. To tackle these issues, this study designs a novel adaptive proportional-integral-derivative (PID) coupling tracking control method with an automatic gain adjustment mechanism, achieving real-time and accurate motion mapping from the physical to the virtual DPC and effectively suppressing the double-pendulum swing, which is a key goal. Particularly, to effectively compensate for different lumped disturbances in both the physical and virtual DPC and improve tracking accuracy, a time-delay control strategy is integrated into the underactuated DPC system, and an adaptive method is employed to estimate the influence of state-dependent time-delay errors online, thereby relaxing conventional boundedness assumptions. Finally, synchronous motion experiments demonstrate the effectiveness and superiority of the proposed virtual-physical tracking control method.
- Open Access
Robust Autonomous Perception for Indoor Service Machines via Geometry-Aware RGB-D SLAM and Probabilistic Dynamic Modeling
Reliable autonomous perception is essential for indoor service machines operating in human-centered environments, where weak textures, repetitive structures, and frequent dynamic interference often degrade localization stability. Conventional RGB-D SLAM systems typically rely on static-scene assumptions or binary semantic masking, which are insufficient for handling persistent and fine-grained environmental dynamics. This paper presents a robust autonomous perception framework based on geometry-aware RGB-D SLAM, with a particular emphasis on probabilistic dynamic modeling at the feature level. The proposed system integrates multi-granularity geometric representations, including point features, parallel-line structures, and planar regions, to enhance geometric observability in low-texture indoor environments. On this basis, a probabilistic dynamic model is introduced to explicitly characterize feature reliability under motion, where dynamic probabilities are initialized by object detection and continuously updated through temporal consistency, spatial propagation, and multi-view geometric verification. Large-scale planar structures further serve as stable anchors to support robust pose estimation. Experimental results on the TUM RGB-D dynamic benchmark demonstrate that the proposed method significantly improves localization robustness, reducing the average ATE RMSE by approximately 66% compared with representative dynamic SLAM baselines. Additional evaluations on a real-world indoor dataset further validate its effectiveness for long-term autonomous perception under dense motion and frequent occlusions.
Open AccessRobust Autonomous Perception for Indoor Service Machines via Geometry-Aware RGB-D SLAM and Probabilistic Dynamic Modeling
Reliable autonomous perception is essential for indoor service machines operating in human-centered environments, where weak textures, repetitive structures, and frequent dynamic interference often degrade localization stability. Conventional RGB-D SLAM systems typically rely on static-scene assumptions or binary semantic masking, which are insufficient for handling persistent and fine-grained environmental dynamics. This paper presents a robust autonomous perception framework based on geometry-aware RGB-D SLAM, with a particular emphasis on probabilistic dynamic modeling at the feature level. The proposed system integrates multi-granularity geometric representations, including point features, parallel-line structures, and planar regions, to enhance geometric observability in low-texture indoor environments. On this basis, a probabilistic dynamic model is introduced to explicitly characterize feature reliability under motion, where dynamic probabilities are initialized by object detection and continuously updated through temporal consistency, spatial propagation, and multi-view geometric verification. Large-scale planar structures further serve as stable anchors to support robust pose estimation. Experimental results on the TUM RGB-D dynamic benchmark demonstrate that the proposed method significantly improves localization robustness, reducing the average ATE RMSE by approximately 66% compared with representative dynamic SLAM baselines. Additional evaluations on a real-world indoor dataset further validate its effectiveness for long-term autonomous perception under dense motion and frequent occlusions.
- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 56 (5) , 3060-3074
Cooperative Output Regulation of Heterogeneous Systems With Actuator Faults: A Double-Layer Event-Triggered Approach
This work addresses the cooperative output regulation (COR) problem of heterogeneous multiagent systems (MASs) subject to actuator faults. First, a novel double-layer event-triggering mechanism (DLETM) is proposed, with appropriately constructed first-layer and second-layer triggering conditions, and allows communication events only at the second-layer’s triggering instants; it ensures that the communication frequency is reduced compared with the existing single-layer event-triggering mechanism (ETM). Then, a double-layer event-triggered (DLET) observer is designed to estimate the exosystem’s matrix with explicit minimum triggering interval (MTI), and an adaptive DLET observer is further developed to estimate the exosystem’s state, such that strong Zeno-freeness is ensured while global topology information is not required to implement the DLET observer. The introduction of the second-layer ETM brings a challenge for stability analysis and DLETM design. To solve this issue, a novel anti-trigonometric function term is added into Lyapunov function with well-designed internal adaptive variables. Moreover, an adaptive control law is designed to handle actuator faults, which ensures that the regulated output converges to zero, whereas most existing methods only ensure bounded error. Finally, simulation results confirm the effectiveness of the proposed approach.
- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 56 (5) , 3060-3074
Cooperative Output Regulation of Heterogeneous Systems With Actuator Faults: A Double-Layer Event-Triggered Approach
This work addresses the cooperative output regulation (COR) problem of heterogeneous multiagent systems (MASs) subject to actuator faults. First, a novel double-layer event-triggering mechanism (DLETM) is proposed, with appropriately constructed first-layer and second-layer triggering conditions, and allows communication events only at the second-layer’s triggering instants; it ensures that the communication frequency is reduced compared with the existing single-layer event-triggering mechanism (ETM). Then, a double-layer event-triggered (DLET) observer is designed to estimate the exosystem’s matrix with explicit minimum triggering interval (MTI), and an adaptive DLET observer is further developed to estimate the exosystem’s state, such that strong Zeno-freeness is ensured while global topology information is not required to implement the DLET observer. The introduction of the second-layer ETM brings a challenge for stability analysis and DLETM design. To solve this issue, a novel anti-trigonometric function term is added into Lyapunov function with well-designed internal adaptive variables. Moreover, an adaptive control law is designed to handle actuator faults, which ensures that the regulated output converges to zero, whereas most existing methods only ensure bounded error. Finally, simulation results confirm the effectiveness of the proposed approach.
- 1 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 56 (2) , 940-953
Data-Driven Event-Triggered Control of Multiagent Systems With Communication Delays: A Hybrid System Approach
This work studies the problem of data-driven event-triggered control of continuous-time multiagent systems (MASs) with communication delays. A hybrid system approach is proposed to address this problem, hybrid dynamic event-triggering mechanism (DETM) is utilized to ensure strong Zeno-freeness and estimated system matrix is introduced to design time-varying state estimator. By using several auxiliary variables, the closed-loop MAS is described into hybrid system form, that can completely describe the flow dynamics and jump dynamics of the MAS in the presence of communication delays. Then, data-based controller gain matrix design is developed and an estimator is designed to estimate unknown system matrix. After that, based on derived model-based stability conditions and combined with data-driven representation of MAS, data-based stability analysis and event-triggering mechanism (ETM) design results are obtained. The main advantages of the proposed approach in contrast with existing works are that accurate system model is not needed and strong Zeno-freeness is ensured under the scenario with communication delays. Finally, the effectiveness of the proposed method is verified by simulation example.
- 1 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Cybernetics
- Vol. 56 (2) , 940-953
Data-Driven Event-Triggered Control of Multiagent Systems With Communication Delays: A Hybrid System Approach
This work studies the problem of data-driven event-triggered control of continuous-time multiagent systems (MASs) with communication delays. A hybrid system approach is proposed to address this problem, hybrid dynamic event-triggering mechanism (DETM) is utilized to ensure strong Zeno-freeness and estimated system matrix is introduced to design time-varying state estimator. By using several auxiliary variables, the closed-loop MAS is described into hybrid system form, that can completely describe the flow dynamics and jump dynamics of the MAS in the presence of communication delays. Then, data-based controller gain matrix design is developed and an estimator is designed to estimate unknown system matrix. After that, based on derived model-based stability conditions and combined with data-driven representation of MAS, data-based stability analysis and event-triggering mechanism (ETM) design results are obtained. The main advantages of the proposed approach in contrast with existing works are that accurate system model is not needed and strong Zeno-freeness is ensured under the scenario with communication delays. Finally, the effectiveness of the proposed method is verified by simulation example.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 23, 14079-14092
Direct Data-Driven Event-Triggered Consensus Under Disturbances and Digraph
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 23, 14079-14092
Direct Data-Driven Event-Triggered Consensus Under Disturbances and Digraph
- 19 December 2025
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (1) , 016305
Visual-IMU fusion for 3D pose measurement of articulated construction equipment
Accurate 3D pose measurement of articulated construction equipment face significant challenges due to the cumulative effect of joint angle errors along the kinematic chain. In systems such as excavators, small angular deviations from onboard inertial measurement units (IMUs) can propagate through forward kinematics and lead to large pose inaccuracies at the end-effector. This study presents a marker-free, multi-sensor measurement framework that integrates IMU-aided joint initialization, monocular visual feedback refinement, and inverse kinematics-based compensation. The system iteratively corrects coarse poses derived from joint angles using real-time image-based tracking and further mitigates residual error by feedback through the kinematic model. Since ground-truth 3D orientation of the excavator bucket is not directly measurable in field conditions, we validate pose measurement using AR-based overlay visualization and quantitative 3D position error analysis. Results from real digging tasks on an excavator platform show stable tracking under occlusion and lighting variation, with end-effector position errors consistently below 0.1 m. These findings demonstrate the effectiveness and applicability of the proposed system for precision pose measurement in articulated heavy machinery.
- 19 December 2025
- Published by IOP Publishing in Measurement Science and Technology
- Vol. 37 (1) , 016305
Visual-IMU fusion for 3D pose measurement of articulated construction equipment
Accurate 3D pose measurement of articulated construction equipment face significant challenges due to the cumulative effect of joint angle errors along the kinematic chain. In systems such as excavators, small angular deviations from onboard inertial measurement units (IMUs) can propagate through forward kinematics and lead to large pose inaccuracies at the end-effector. This study presents a marker-free, multi-sensor measurement framework that integrates IMU-aided joint initialization, monocular visual feedback refinement, and inverse kinematics-based compensation. The system iteratively corrects coarse poses derived from joint angles using real-time image-based tracking and further mitigates residual error by feedback through the kinematic model. Since ground-truth 3D orientation of the excavator bucket is not directly measurable in field conditions, we validate pose measurement using AR-based overlay visualization and quantitative 3D position error analysis. Results from real digging tasks on an excavator platform show stable tracking under occlusion and lighting variation, with end-effector position errors consistently below 0.1 m. These findings demonstrate the effectiveness and applicability of the proposed system for precision pose measurement in articulated heavy machinery.
- 2 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 36 (4) , 4900-4912
CL-WTAL: Weakly-Supervised Temporal Complex Action Localization Based on Multi-Scale Contrast Learning
Temporal action localization in long-term untrimmed videos remains a critical yet challenging task in video understanding, with existing methods often relying on anchor-based or fully-supervised frameworks that incur heavy computation and require labor-intensive frame-level annotations. This paper presents a novel weakly-supervised approach, CL-WTAL, which leverages multi-scale contrast learning and graph convolution for accurate action localization and recognition. The method comprises three key components: (i) A multi-scale sliding window mechanism (long/normal/short sequences) to segment sub-actions from complex videos, adapting to diverse action durations; (ii) A spatio-temporal graph convolution network (ST-RGCN) to extract skeletal feature vectors, integrating human motion dynamics and environmental context; (iii) A contrastive learning-based similarity evaluation framework that combines cosine similarity and Dynamic Time Warping (DTW) distance to measure feature vector relationships, enabling precise action boundary detection without extensive fine-tuning. Experiments on daily-life video datasets demonstrate that CL-WTAL effectively localizes action intervals and classifies actions with high accuracy, outperforming state-of-the-art weakly-supervised methods.
- 2 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems for Video Technology
- Vol. 36 (4) , 4900-4912
CL-WTAL: Weakly-Supervised Temporal Complex Action Localization Based on Multi-Scale Contrast Learning
Temporal action localization in long-term untrimmed videos remains a critical yet challenging task in video understanding, with existing methods often relying on anchor-based or fully-supervised frameworks that incur heavy computation and require labor-intensive frame-level annotations. This paper presents a novel weakly-supervised approach, CL-WTAL, which leverages multi-scale contrast learning and graph convolution for accurate action localization and recognition. The method comprises three key components: (i) A multi-scale sliding window mechanism (long/normal/short sequences) to segment sub-actions from complex videos, adapting to diverse action durations; (ii) A spatio-temporal graph convolution network (ST-RGCN) to extract skeletal feature vectors, integrating human motion dynamics and environmental context; (iii) A contrastive learning-based similarity evaluation framework that combines cosine similarity and Dynamic Time Warping (DTW) distance to measure feature vector relationships, enabling precise action boundary detection without extensive fine-tuning. Experiments on daily-life video datasets demonstrate that CL-WTAL effectively localizes action intervals and classifies actions with high accuracy, outperforming state-of-the-art weakly-supervised methods.