Xiangfei Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Xiangfei Li has a total of 42 co-authors. Here are the top co-authors by number of shared publications:
- Huan Zhao
Huazhong University of Science and Technology
19 shared publications Han Ding
15 shared publications- Han Ding
Huazhong University of Science and Technology
11 shared publications Huan Zhao
7 shared publicationsXin Zhao
Huazhong Univ Sci & Technol, State Key Lab Digital Mfg Equipment & Technol, Wuhan 430074, Hubei, Peoples R China
5 shared publicationsYecan Yin
3 shared publications
Latest Publications
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (10) , 12288-12295
Physics-Informed Sliding-Window Particle Filtering for Tactile-Only In-Hand 6-DoF Object Pose Refinement
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (10) , 12288-12295
Physics-Informed Sliding-Window Particle Filtering for Tactile-Only In-Hand 6-DoF Object Pose Refinement
- 25 June 2026
- Published by Elsevier in Journal of Manufacturing Systems
- Vol. 88, 350-375
A comprehensive review of robotic assembly of aerospace components
- 25 June 2026
- Published by Elsevier in Journal of Manufacturing Systems
- Vol. 88, 350-375
A comprehensive review of robotic assembly of aerospace components
- 4 May 2026
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 203
Decoupled image-based visual servoing with visibility and finite-time convergence performance constraints
- 4 May 2026
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 203
Decoupled image-based visual servoing with visibility and finite-time convergence performance constraints
- 11 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 23996-24008
Virtual Image-Based Visual Servoing
The classical image-based visual servoing (IBVS) methods exhibit strong robustness to robot modeling and camera calibration errors, but suffer from uncontrollable spatial trajectories and local convergence issues. Although current IBVS variants employ specialized visual features or information to improve trajectory controllability, these approaches impose restrictive geometric constraints, such as requiring coplanar feature points or configurations approximately orthogonal to the optical axis of the camera. Furthermore, despite conclusive evidence of local minima (LM) in the IBVS scheme, there is currently no solution to address this challenge. For the reason, this article first introduces a novel virtual image construction method and proposes a decoupling visual servoing control law based on virtual image, achieving explicit separation of translational and rotational error dynamics. Then, through Monte Carlo simulations, the spatial distribution patterns of local minima in IBVS are systematically studied, and an escape algorithm leveraging virtual image is further designed. To the best of our knowledge, this may be the first attempt to address the issue of local minima only through IBVS to a certain extent. Comparative simulations and experiments validate the effectiveness and superiority of the proposed decoupling control law based on virtual image, and the success rate of the local minima escape algorithm is 100% under different configurations.
- 11 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 23996-24008
Virtual Image-Based Visual Servoing
The classical image-based visual servoing (IBVS) methods exhibit strong robustness to robot modeling and camera calibration errors, but suffer from uncontrollable spatial trajectories and local convergence issues. Although current IBVS variants employ specialized visual features or information to improve trajectory controllability, these approaches impose restrictive geometric constraints, such as requiring coplanar feature points or configurations approximately orthogonal to the optical axis of the camera. Furthermore, despite conclusive evidence of local minima (LM) in the IBVS scheme, there is currently no solution to address this challenge. For the reason, this article first introduces a novel virtual image construction method and proposes a decoupling visual servoing control law based on virtual image, achieving explicit separation of translational and rotational error dynamics. Then, through Monte Carlo simulations, the spatial distribution patterns of local minima in IBVS are systematically studied, and an escape algorithm leveraging virtual image is further designed. To the best of our knowledge, this may be the first attempt to address the issue of local minima only through IBVS to a certain extent. Comparative simulations and experiments validate the effectiveness and superiority of the proposed decoupling control law based on virtual image, and the success rate of the local minima escape algorithm is 100% under different configurations.
- 12 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 31 (2) , 1357-1368
Macro–Micro Composite Robot Precision Assembly by a Novel 2-D Active End-Effector
Robotic positioning accuracy limitations make it difficult to achieve precise hole-searching and positioning in high-precision assembly tasks, hindering successful assembly completion. To overcome this challenge, this article proposes a macro–micro composite precision assembly method that integrates a robot with a 2-D active end-effector (TDAE). In this method, the robot provides coarse positioning in a large workspace, while the end-effector handles high-precision positioning within a confined range. To the best of the authors’ knowledge, this may be the first attempt to realize precise peg-in-hole assembly by combining the robot and active end-effector. To ensure high-precision control of the TDAE, a hybrid control strategy incorporating nominal control, disturbance observation, and feedforward compensation is proposed. Controller parameter constraint intervals are established based on the accuracy and stability requirements of the actuator. Experimental results show that the proposed TDAE achieves high-precision positioning with an error margin of 5 $\mu$m within a $\pm$ 3 mm range. The $x$-axis and $y$-axis bandwidths reach 145 and 200 Hz, respectively. Comparative experiments with a robotic hole-searching process further verify the system’s stability and efficiency, ensuring smooth execution of the automated robotic assembly process.
- 12 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 31 (2) , 1357-1368
Macro–Micro Composite Robot Precision Assembly by a Novel 2-D Active End-Effector
Robotic positioning accuracy limitations make it difficult to achieve precise hole-searching and positioning in high-precision assembly tasks, hindering successful assembly completion. To overcome this challenge, this article proposes a macro–micro composite precision assembly method that integrates a robot with a 2-D active end-effector (TDAE). In this method, the robot provides coarse positioning in a large workspace, while the end-effector handles high-precision positioning within a confined range. To the best of the authors’ knowledge, this may be the first attempt to realize precise peg-in-hole assembly by combining the robot and active end-effector. To ensure high-precision control of the TDAE, a hybrid control strategy incorporating nominal control, disturbance observation, and feedforward compensation is proposed. Controller parameter constraint intervals are established based on the accuracy and stability requirements of the actuator. Experimental results show that the proposed TDAE achieves high-precision positioning with an error margin of 5 $\mu$m within a $\pm$ 3 mm range. The $x$-axis and $y$-axis bandwidths reach 145 and 200 Hz, respectively. Comparative experiments with a robotic hole-searching process further verify the system’s stability and efficiency, ensuring smooth execution of the automated robotic assembly process.
- Open Access
- 11 June 2025
- Published by Springer Nature in Journal of Intelligent & Robotic Systems
- Vol. 111 (3) , 1-19
Event-Based Collaborative Planning and Control with Loosely-Coupled Transportation Model for Dual Mobile Manipulators
- 11 June 2025
- Published by Springer Nature in Journal of Intelligent & Robotic Systems
- Vol. 111 (3) , 1-19
Open AccessEvent-Based Collaborative Planning and Control with Loosely-Coupled Transportation Model for Dual Mobile Manipulators
- 22 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 15792-15804
A Novel Deep Reinforcement Learning-Based Path/Force Cooperative Regulation Framework for Dual-Arm Object Transportation
Dual-arm robots, with their high flexibility and broad operational range, are widely used in industrial and household transportation tasks. However, their high degrees of motion freedom and the closed-chain constraints formed during transportation pose challenges for path planning and force control. This study proposes a dual-agent control framework based on the Soft-Actor-Critic (SAC) algorithm of Deep Reinforcement Learning (DRL), where one agent is responsible for path planning and the other handles force control. This framework enables dual-arm robots to achieve dynamic obstacle avoidance, avoid unsolvable configurations and singularities, and generate efficient and smooth paths, while also fulfilling internal force tracking requirements based on task demands. Additionally, it addresses the gap in transferring the force control agent from simulation to real-world applications through a state mapping network, and the force control agent does not require retraining for different objects. Finally, the effectiveness of the proposed framework is validated through two scenes, including multiple bookshelves stacking and dynamic obstacle avoidance during the box transportation. Validation was also carried out with objects of different geometries and weights. Note to Practitioners—The motivation of this study is the path planning and force tracking control problem for dual-arm transportation. Existing methods rarely provide a comprehensive framework to address the task of dual-arm transporting different objects in dynamic environments. A DRL-based dual-agent control framework is proposed in this study, which simultaneously achieves task-compliant path planning and force tracking control. The trained agents exhibit strong generalization capabilities and stability, allowing them to handle objects of different geometries, weights, and stiffness without retraining. The limitation lies in the lack of methods for generating dual-arm initial grasping actions for complex-shaped objects. Future research will integrate image recognition techniques to enable grasping action generation for objects in diverse poses, further enhancing the applicability of the proposed framework.
- 22 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 15792-15804
A Novel Deep Reinforcement Learning-Based Path/Force Cooperative Regulation Framework for Dual-Arm Object Transportation
Dual-arm robots, with their high flexibility and broad operational range, are widely used in industrial and household transportation tasks. However, their high degrees of motion freedom and the closed-chain constraints formed during transportation pose challenges for path planning and force control. This study proposes a dual-agent control framework based on the Soft-Actor-Critic (SAC) algorithm of Deep Reinforcement Learning (DRL), where one agent is responsible for path planning and the other handles force control. This framework enables dual-arm robots to achieve dynamic obstacle avoidance, avoid unsolvable configurations and singularities, and generate efficient and smooth paths, while also fulfilling internal force tracking requirements based on task demands. Additionally, it addresses the gap in transferring the force control agent from simulation to real-world applications through a state mapping network, and the force control agent does not require retraining for different objects. Finally, the effectiveness of the proposed framework is validated through two scenes, including multiple bookshelves stacking and dynamic obstacle avoidance during the box transportation. Validation was also carried out with objects of different geometries and weights. Note to Practitioners—The motivation of this study is the path planning and force tracking control problem for dual-arm transportation. Existing methods rarely provide a comprehensive framework to address the task of dual-arm transporting different objects in dynamic environments. A DRL-based dual-agent control framework is proposed in this study, which simultaneously achieves task-compliant path planning and force tracking control. The trained agents exhibit strong generalization capabilities and stability, allowing them to handle objects of different geometries, weights, and stiffness without retraining. The limitation lies in the lack of methods for generating dual-arm initial grasping actions for complex-shaped objects. Future research will integrate image recognition techniques to enable grasping action generation for objects in diverse poses, further enhancing the applicability of the proposed framework.
- 16 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 30 (6) , 5928-5939
Robotic Grinding Skills Learning Based on Geodesic Length Dynamic Motion Primitives
Learning grinding skills from human craftsmen by imitation learning has emerged as a prominent research topic in the field of robotic machining. Given their robust trajectory generalization ability and resilience to various external disturbances and environmental changes, dynamical movement primitives (DMPs) provide a promising skills learning solution for the robotic grinding. However, challenges arise when directly applying DMPs to grinding tasks, including low orientation accuracy, inaccurate synchronization of position, orientation, and force, and the inability to generalize surface trajectories. To address these issues, this article proposes a robotic grinding skills learning method based on geodesic length DMPs (Geo-DMPs). First, a normalized 2-D weighted Gaussian kernel function and intrinsic mean clustering algorithm are proposed to extract surface geometric features from multiple demonstration trajectories. Then, an orientation manifold distance metric is introduced to exclude the time factor from the classical orientation DMPs, thereby constructing Geo-DMPs for the orientation learning to improve the orientation trajectory generation accuracy. On this basis, a synchronization encoding framework for position, orientation, and force skills is established, using a phase function related to geodesic length. This framework enables the generation of robotic grinding actions between any two points on the surface. Finally, experiments on robotic chamfer grinding and free-form surface grinding demonstrate that the proposed method exhibits high geometric accuracy and good generalization capabilities in encoding and generating grinding skills. This method holds significant implications for learning and promoting robotic grinding skills. To the best of our knowledge, this may be the first attempt to use DMPs to generate grinding skills for position, orientation, and force on model-free surfaces, thereby presenting a novel approach to robotic grinding skills learning.
- 16 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 30 (6) , 5928-5939
Robotic Grinding Skills Learning Based on Geodesic Length Dynamic Motion Primitives
Learning grinding skills from human craftsmen by imitation learning has emerged as a prominent research topic in the field of robotic machining. Given their robust trajectory generalization ability and resilience to various external disturbances and environmental changes, dynamical movement primitives (DMPs) provide a promising skills learning solution for the robotic grinding. However, challenges arise when directly applying DMPs to grinding tasks, including low orientation accuracy, inaccurate synchronization of position, orientation, and force, and the inability to generalize surface trajectories. To address these issues, this article proposes a robotic grinding skills learning method based on geodesic length DMPs (Geo-DMPs). First, a normalized 2-D weighted Gaussian kernel function and intrinsic mean clustering algorithm are proposed to extract surface geometric features from multiple demonstration trajectories. Then, an orientation manifold distance metric is introduced to exclude the time factor from the classical orientation DMPs, thereby constructing Geo-DMPs for the orientation learning to improve the orientation trajectory generation accuracy. On this basis, a synchronization encoding framework for position, orientation, and force skills is established, using a phase function related to geodesic length. This framework enables the generation of robotic grinding actions between any two points on the surface. Finally, experiments on robotic chamfer grinding and free-form surface grinding demonstrate that the proposed method exhibits high geometric accuracy and good generalization capabilities in encoding and generating grinding skills. This method holds significant implications for learning and promoting robotic grinding skills. To the best of our knowledge, this may be the first attempt to use DMPs to generate grinding skills for position, orientation, and force on model-free surfaces, thereby presenting a novel approach to robotic grinding skills learning.
- Open Access
Robotic Grinding Skills Learning Based on Geodesic Length Dynamic Motion Primitives
Learning grinding skills from human craftsmen via imitation learning has become a key research topic in robotic machining. Due to their strong generalization and robustness to external disturbances, Dynamical Movement Primitives (DMPs) offer a promising approach for robotic grinding skill learning. However, directly applying DMPs to grinding tasks faces challenges, such as low orientation accuracy, unsynchronized position-orientation-force, and limited generalization for surface trajectories. To address these issues, this paper proposes a robotic grinding skill learning method based on geodesic length DMPs (Geo-DMPs). First, a normalized 2D weighted Gaussian kernel and intrinsic mean clustering algorithm are developed to extract geometric features from multiple demonstrations. Then, an orientation manifold distance metric removes the time dependency in traditional orientation DMPs, enabling accurate orientation learning via Geo-DMPs. A synchronization encoding framework is further proposed to jointly model position, orientation, and force using a geodesic length-based phase function. This framework enables robotic grinding actions to be generated between any two surface points. Experiments on robotic chamfer grinding and free-form surface grinding validate that the proposed method achieves high geometric accuracy and generalization in skill encoding and generation. To our knowledge, this is the first attempt to use DMPs for jointly learning and generating grinding skills in position, orientation, and force on model-free surfaces, offering a novel path for robotic grinding.
Open AccessRobotic Grinding Skills Learning Based on Geodesic Length Dynamic Motion Primitives
Learning grinding skills from human craftsmen via imitation learning has become a key research topic in robotic machining. Due to their strong generalization and robustness to external disturbances, Dynamical Movement Primitives (DMPs) offer a promising approach for robotic grinding skill learning. However, directly applying DMPs to grinding tasks faces challenges, such as low orientation accuracy, unsynchronized position-orientation-force, and limited generalization for surface trajectories. To address these issues, this paper proposes a robotic grinding skill learning method based on geodesic length DMPs (Geo-DMPs). First, a normalized 2D weighted Gaussian kernel and intrinsic mean clustering algorithm are developed to extract geometric features from multiple demonstrations. Then, an orientation manifold distance metric removes the time dependency in traditional orientation DMPs, enabling accurate orientation learning via Geo-DMPs. A synchronization encoding framework is further proposed to jointly model position, orientation, and force using a geodesic length-based phase function. This framework enables robotic grinding actions to be generated between any two surface points. Experiments on robotic chamfer grinding and free-form surface grinding validate that the proposed method achieves high geometric accuracy and generalization in skill encoding and generation. To our knowledge, this is the first attempt to use DMPs for jointly learning and generating grinding skills in position, orientation, and force on model-free surfaces, offering a novel path for robotic grinding.
- 21 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 41 (15523098) , 1456-1475
Industrial Robots Energy Consumption Modeling, Identification and Optimization Through Time-Scaling
Industrial robots (IRs) have considerable energy-saving potential due to their vast application scale and wide range of applications. Although substantial work on the energy consumption (EC) optimization of IRs has emerged, most optimization approaches require prior knowledge of the IRs' dynamic characteristics and the electro-mechanical parameters of their drive systems, which are typically not provided by IR manufacturers. Therefore, this paper proposes an EC modeling and optimization method based on the time-scaling technique and custom identification experimental data without joint torque information. Specifically, this paper develops an energy characteristic parameter sub-model (ECPSM) to formulate the EC resulting from configuration transitions. Additionally, theoretical proof demonstrates that all coefficients in the proposed ECPSM can be identified based on the data of a finite number of identification experiments. Building upon the proposed EC model, a bi-directional dynamic programming (BDP) algorithm optimizes the IR's trajectory for energy-saving, while utilizing parallel processing significantly reduces the time required for the optimization process. Experimental results on the KUKA KR60-3 demonstrate that the proposed method achieves an average relative error of 1.59% for predicting the EC of linear scaling trajectories and 6.19% for nonlinear scaled trajectories. Moreover, the BDP-based optimization method dramatically reduces the computational time required to obtain the optimal scaling trajectory and its EC.
- 21 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Robotics
- Vol. 41 (15523098) , 1456-1475
Industrial Robots Energy Consumption Modeling, Identification and Optimization Through Time-Scaling
Industrial robots (IRs) have considerable energy-saving potential due to their vast application scale and wide range of applications. Although substantial work on the energy consumption (EC) optimization of IRs has emerged, most optimization approaches require prior knowledge of the IRs' dynamic characteristics and the electro-mechanical parameters of their drive systems, which are typically not provided by IR manufacturers. Therefore, this paper proposes an EC modeling and optimization method based on the time-scaling technique and custom identification experimental data without joint torque information. Specifically, this paper develops an energy characteristic parameter sub-model (ECPSM) to formulate the EC resulting from configuration transitions. Additionally, theoretical proof demonstrates that all coefficients in the proposed ECPSM can be identified based on the data of a finite number of identification experiments. Building upon the proposed EC model, a bi-directional dynamic programming (BDP) algorithm optimizes the IR's trajectory for energy-saving, while utilizing parallel processing significantly reduces the time required for the optimization process. Experimental results on the KUKA KR60-3 demonstrate that the proposed method achieves an average relative error of 1.59% for predicting the EC of linear scaling trajectories and 6.19% for nonlinear scaled trajectories. Moreover, the BDP-based optimization method dramatically reduces the computational time required to obtain the optimal scaling trajectory and its EC.