Xiaojian 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
Xiaojian Li has a total of 130 co-authors. Here are the top co-authors by number of shared publications:
- Dong Sun
Department of Biomedical Engineering, City University of Hong Kong, Kowloon, Hong Kong, China
7 shared publications - Ling Li
School of Management and the Key Laboratory of Process Optimization and Intelligent Decision-Making (Ministry of Education), Hefei University of Technology, Hefei, China
7 shared publications - Shanlin Yang
Hefei University of Technology
7 shared publications - Shuai Ding
Hefei University of Technology
6 shared publications - Hangjie Mo
School of Management, Hefei University of Technology, Hefei, China
5 shared publications Hangjie Mo
School of Management and the Key Laboratory of Process Optimization and Intelligent Decision-Making (Ministry of Education), Hefei University of Technology, Hefei, China
3 shared publications
Latest Publications
- 2 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Robotics and Bionics
- Vol. PP (3) , 1
Design and Evaluation of a Compact, MR-Compatible Origami Continuum Robot with Variable Length
Continuum robots have recently gained attention for their great value in medical applications. This paper presents a novel design approach for MR-compatible medical continuum robots with the ability to variable length and controllable motion at compact size and light weight. The robot employs a small-scale origami continuum structure with an outside diameter of 15 mm, is driven by three tendon-shape memory alloy (SMA) actuators, and senses the motion of actuators by a thin linear potentiometer (Thinpot). The robot system is lightweight at 76.7 g and compact, with the proximal sensing and actuation module measuring only 34 mm in diameter and 90 mm in length. It is first demonstrated that the Thinpot is insensitive to magnetic disturbance and MR-compatible and can realize accurate feedback control of the large stroke of SMA actuators, even in a real 1.5T MR environment. Further, the bending angle and tip position of the robot can be controlled based on an inverse kinematic model. The robots equipped with different tools will show versatility in biomedical applications, such as endoscopy and needle insertion.
- 2 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Robotics and Bionics
- Vol. PP (3) , 1
Design and Evaluation of a Compact, MR-Compatible Origami Continuum Robot with Variable Length
Continuum robots have recently gained attention for their great value in medical applications. This paper presents a novel design approach for MR-compatible medical continuum robots with the ability to variable length and controllable motion at compact size and light weight. The robot employs a small-scale origami continuum structure with an outside diameter of 15 mm, is driven by three tendon-shape memory alloy (SMA) actuators, and senses the motion of actuators by a thin linear potentiometer (Thinpot). The robot system is lightweight at 76.7 g and compact, with the proximal sensing and actuation module measuring only 34 mm in diameter and 90 mm in length. It is first demonstrated that the Thinpot is insensitive to magnetic disturbance and MR-compatible and can realize accurate feedback control of the large stroke of SMA actuators, even in a real 1.5T MR environment. Further, the bending angle and tip position of the robot can be controlled based on an inverse kinematic model. The robots equipped with different tools will show versatility in biomedical applications, such as endoscopy and needle insertion.
- 13 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (5) , 7249-7260
Adaptive Deep Koopman Operators for Soft Robot Control: Application to Luban Lock Disassembly
Soft robots have attracted considerable attention due to their intrinsic compliance and environmental adaptability in complex dynamic scenarios. However, their infinite degrees of freedom and strongly nonlinear dynamics present fundamental challenges in precise modeling and control. To address this issue, this article proposes an adaptive deep Koopman (ADK) modeling approach for model-based optimal control of soft robots. Our approach incorporates the control-oriented loss functions into the deep Koopman modeling framework to regulate open-loop poles and singular values of the controllability matrix, bridging the critical gap between the Koopman model learning and closed-loop control. The model parameters in the Koopman observable space are further updated online to account for dynamic changes under material wear and load variations. We develop a model-based optimal controller with the ADK model for soft robots, where an integral action is introduced to improve the control performance under varying load conditions and inaccurate feedforward control inputs. Extensive experimental validations including set-point regulation under varying load conditions and autonomous disassembly of a three-way Luban lock are conducted, which demonstrate the effectiveness and robustness of our approach.
- 13 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (5) , 7249-7260
Adaptive Deep Koopman Operators for Soft Robot Control: Application to Luban Lock Disassembly
Soft robots have attracted considerable attention due to their intrinsic compliance and environmental adaptability in complex dynamic scenarios. However, their infinite degrees of freedom and strongly nonlinear dynamics present fundamental challenges in precise modeling and control. To address this issue, this article proposes an adaptive deep Koopman (ADK) modeling approach for model-based optimal control of soft robots. Our approach incorporates the control-oriented loss functions into the deep Koopman modeling framework to regulate open-loop poles and singular values of the controllability matrix, bridging the critical gap between the Koopman model learning and closed-loop control. The model parameters in the Koopman observable space are further updated online to account for dynamic changes under material wear and load variations. We develop a model-based optimal controller with the ADK model for soft robots, where an integral action is introduced to improve the control performance under varying load conditions and inaccurate feedforward control inputs. Extensive experimental validations including set-point regulation under varying load conditions and autonomous disassembly of a three-way Luban lock are conducted, which demonstrate the effectiveness and robustness of our approach.
- Open Access
From EMR Data to Clinical Insight: An LLM-Driven Framework for Automated Pre-Consultation Questionnaire Generation
Pre-consultation is a critical component of effective healthcare delivery. However, generating comprehensive pre-consultation questionnaires from complex, voluminous Electronic Medical Records (EMRs) is a challenging task. Direct Large Language Model (LLM) approaches face difficulties in this task, particularly regarding information completeness, logical order, and disease-level synthesis. To address this issue, we propose a novel multi-stage LLM-driven framework: Stage 1 extracts atomic assertions (key facts with timing) from EMRs; Stage 2 constructs personal causal networks and synthesizes disease knowledge by clustering representative networks from an EMR corpus; Stage 3 generates tailored personal and standardized disease-specific questionnaires based on these structured representations. This framework overcomes limitations of direct methods by building explicit clinical knowledge. Evaluated on a real-world EMR dataset and validated by clinical experts, our method demonstrates superior performance in information coverage, diagnostic relevance, understandability, and generation time, highlighting its practical potential to enhance patient information collection.
Open AccessFrom EMR Data to Clinical Insight: An LLM-Driven Framework for Automated Pre-Consultation Questionnaire Generation
Pre-consultation is a critical component of effective healthcare delivery. However, generating comprehensive pre-consultation questionnaires from complex, voluminous Electronic Medical Records (EMRs) is a challenging task. Direct Large Language Model (LLM) approaches face difficulties in this task, particularly regarding information completeness, logical order, and disease-level synthesis. To address this issue, we propose a novel multi-stage LLM-driven framework: Stage 1 extracts atomic assertions (key facts with timing) from EMRs; Stage 2 constructs personal causal networks and synthesizes disease knowledge by clustering representative networks from an EMR corpus; Stage 3 generates tailored personal and standardized disease-specific questionnaires based on these structured representations. This framework overcomes limitations of direct methods by building explicit clinical knowledge. Evaluated on a real-world EMR dataset and validated by clinical experts, our method demonstrates superior performance in information coverage, diagnostic relevance, understandability, and generation time, highlighting its practical potential to enhance patient information collection.
- 3 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 55 (10) , 6545-6557
Enhancing Robotic Surgery With Haptic Feedback: A Cooperative Control Strategy for Autonomous Laparoscope Control
The development of autonomous laparoscope control in robot-assisted surgery has emerged as a significant research area, particularly due to its potential to reduce assistant fatigue and minimize miscommunication between the surgeon and assistant. A notable challenge, however, is the tendency of autonomous control strategies to override the surgeon’s direct command occasionally. To address this issue, we propose a novel haptic feedback-based cooperative control strategy that enhances the surgeon’s command of laparoscopic field of view (FOV) movement in robot-assisted laparoscopic surgery. Specifically, we first established a dynamic model of the laparoscope-holding robot, which serves as a link between the movement of the laparoscopic FOV and the surgical instruments to deliver haptic feedback to the surgeon. Next, a motion observer was developed to transform 30 Hz visual feedback into 1 kHz haptic feedback by integrating visual tracking data with kinematic information, ensuring smoother and more continuous haptic feedback. Finally, we propose two distinct collaboration modes: the plane tracking mode (PTM) ensures instruments remain within the laparoscopic image, and the space tracking mode (STM) synchronizes the laparoscope with instrument movement. The laboratory experiments validated the effectiveness of the proposed method in enhancing the cooperative performance of robot-assisted laparoscope systems while animal experiments demonstrated the feasibility of the PTM design.
- 3 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 55 (10) , 6545-6557
Enhancing Robotic Surgery With Haptic Feedback: A Cooperative Control Strategy for Autonomous Laparoscope Control
The development of autonomous laparoscope control in robot-assisted surgery has emerged as a significant research area, particularly due to its potential to reduce assistant fatigue and minimize miscommunication between the surgeon and assistant. A notable challenge, however, is the tendency of autonomous control strategies to override the surgeon’s direct command occasionally. To address this issue, we propose a novel haptic feedback-based cooperative control strategy that enhances the surgeon’s command of laparoscopic field of view (FOV) movement in robot-assisted laparoscopic surgery. Specifically, we first established a dynamic model of the laparoscope-holding robot, which serves as a link between the movement of the laparoscopic FOV and the surgical instruments to deliver haptic feedback to the surgeon. Next, a motion observer was developed to transform 30 Hz visual feedback into 1 kHz haptic feedback by integrating visual tracking data with kinematic information, ensuring smoother and more continuous haptic feedback. Finally, we propose two distinct collaboration modes: the plane tracking mode (PTM) ensures instruments remain within the laparoscopic image, and the space tracking mode (STM) synchronizes the laparoscope with instrument movement. The laboratory experiments validated the effectiveness of the proposed method in enhancing the cooperative performance of robot-assisted laparoscope systems while animal experiments demonstrated the feasibility of the PTM design.
- 23 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 15990-16000
A Safety-Enhanced Multi-Modal Objectives Motion Fusion Method for Autonomous Retraction in Robotic Surgery
In minimally invasive surgery (MIS), tissue retraction is critical for exposing the surgical site and facilitating pathological tissue excision. However, tissue retraction in MIS is subject to multiple constraints, including restricted field of view, non-damaging tissue retraction force, and limited instrument operating range, which hinder the safe and continuous retraction of tissue during robotic-assisted surgery. This paper introduces a novel autonomous retraction method for MIS, capable of safely retracting tissue within the aforementioned multiple constraints, assisting the surgeon in tissue excision. The method takes into account information from three different modalities: vision, force, and position. Based on these, three distinct control objectives are defined: retraction angle, retraction force, and safety space constraints, with an individual controller and model-predictive evaluation function designed for each objective. We propose a novel multi-objective motion fusion strategy designed to balance three distinct control objectives. This strategy evaluates the sensitivity of each objective to changes in the system state by comparing the gradients of their respective prediction evaluation functions and fuses the control inputs of the individual controllers. The proposed method allows for rapid addition or removal of objectives without altering the algorithmic framework. Experiments with phantoms and ex vivo animal tissue are conducted on robotic platform to validate the effectiveness of the proposed method in various configurations. Note to Practitioners—The motivation of this study is to address the challenge of autonomous tissue retraction in robotic-assisted laparoscopic surgery under multiple constraints. Existing methods primarily focus on single objectives and overlook the conflicts and coordination required among multi-modal objectives such as retraction angle, force, and safety space constraints. These limitations reduce the safety of robotic systems in complex surgical scenarios. A safety-enhanced multi-modal objective fusion method was proposed, integrating vision, force, and position information through dynamic weight allocation based on model-predictive mechanism. The method allows flexible addition or removal of control objectives and constraints, ensuring robust performance in varying conditions. This work provides a practical solution to improve the safety of autonomous robotic retraction, offering valuable insights for practitioners aiming to enhance robotic systems in constrained environments.
- 23 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 22 (15455955) , 15990-16000
A Safety-Enhanced Multi-Modal Objectives Motion Fusion Method for Autonomous Retraction in Robotic Surgery
In minimally invasive surgery (MIS), tissue retraction is critical for exposing the surgical site and facilitating pathological tissue excision. However, tissue retraction in MIS is subject to multiple constraints, including restricted field of view, non-damaging tissue retraction force, and limited instrument operating range, which hinder the safe and continuous retraction of tissue during robotic-assisted surgery. This paper introduces a novel autonomous retraction method for MIS, capable of safely retracting tissue within the aforementioned multiple constraints, assisting the surgeon in tissue excision. The method takes into account information from three different modalities: vision, force, and position. Based on these, three distinct control objectives are defined: retraction angle, retraction force, and safety space constraints, with an individual controller and model-predictive evaluation function designed for each objective. We propose a novel multi-objective motion fusion strategy designed to balance three distinct control objectives. This strategy evaluates the sensitivity of each objective to changes in the system state by comparing the gradients of their respective prediction evaluation functions and fuses the control inputs of the individual controllers. The proposed method allows for rapid addition or removal of objectives without altering the algorithmic framework. Experiments with phantoms and ex vivo animal tissue are conducted on robotic platform to validate the effectiveness of the proposed method in various configurations. Note to Practitioners—The motivation of this study is to address the challenge of autonomous tissue retraction in robotic-assisted laparoscopic surgery under multiple constraints. Existing methods primarily focus on single objectives and overlook the conflicts and coordination required among multi-modal objectives such as retraction angle, force, and safety space constraints. These limitations reduce the safety of robotic systems in complex surgical scenarios. A safety-enhanced multi-modal objective fusion method was proposed, integrating vision, force, and position information through dynamic weight allocation based on model-predictive mechanism. The method allows flexible addition or removal of control objectives and constraints, ensuring robust performance in varying conditions. This work provides a practical solution to improve the safety of autonomous robotic retraction, offering valuable insights for practitioners aiming to enhance robotic systems in constrained environments.
- Open Access
- 11 February 2025
- Published by Wiley in Advanced Intelligent Systems
- Vol. 7 (7)
A Shared Control Method of Multiobjective Motion Fusion for Surgical Robot
Shared control combines human intelligence with autonomous algorithms, demonstrating potential for enhancing the intricate and dynamic operational capabilities of robots. In robotic surgery, shared control can integrate autonomous functions into master–slave control, which enables surgeons to receive certain control assistance while maintaining control of the operation. Such a shared control system must handle multiple types of dynamically changing motion tasks, for instance, trajectory tracking, obstacle avoidance, spatial constraint, and safety guarantee. This article proposes an innovative shared control method based on multiobjective motion fusion (MoMF) by integrating multiple objective controllers into master–slave control mode. In addition, a state feedback mechanism with prediction and evaluation model is developed to balance the control inputs of different control objectives. Further, an objective priority partition function is designed to overcome the conflicts between control objectives. In the proposed method, objective controllers can be added or removed as needed. The theoretical stability of the proposed MoMF method is proved. As an example, a fusion controller that considers four control objectives of a surgical cutting operation is designed to demonstrate the practicability of this method. Finally, the feasibility and effectiveness of the proposed MoMF method are validated by simulations and prosthesis experiments.
- 11 February 2025
- Published by Wiley in Advanced Intelligent Systems
- Vol. 7 (7)
Open AccessA Shared Control Method of Multiobjective Motion Fusion for Surgical Robot
Shared control combines human intelligence with autonomous algorithms, demonstrating potential for enhancing the intricate and dynamic operational capabilities of robots. In robotic surgery, shared control can integrate autonomous functions into master–slave control, which enables surgeons to receive certain control assistance while maintaining control of the operation. Such a shared control system must handle multiple types of dynamically changing motion tasks, for instance, trajectory tracking, obstacle avoidance, spatial constraint, and safety guarantee. This article proposes an innovative shared control method based on multiobjective motion fusion (MoMF) by integrating multiple objective controllers into master–slave control mode. In addition, a state feedback mechanism with prediction and evaluation model is developed to balance the control inputs of different control objectives. Further, an objective priority partition function is designed to overcome the conflicts between control objectives. In the proposed method, objective controllers can be added or removed as needed. The theoretical stability of the proposed MoMF method is proved. As an example, a fusion controller that considers four control objectives of a surgical cutting operation is designed to demonstrate the practicability of this method. Finally, the feasibility and effectiveness of the proposed MoMF method are validated by simulations and prosthesis experiments.
- 3 October 2024
- Published by Springer Nature
- p. 391-400
Misaligned 3D Texture Optimization in MIS Utilizing Generative Framework
- 3 October 2024
- Published by Springer Nature
- p. 391-400
Misaligned 3D Texture Optimization in MIS Utilizing Generative Framework
- Open Access
- 9 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 731-737
Automated Cross-modal Registration Algorithm For Liver Surgery Navigation
- 9 August 2024
- Published by Association for Computing Machinery (ACM)
- p. 731-737
Open AccessAutomated Cross-modal Registration Algorithm For Liver Surgery Navigation
- 5 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-13
Real-Time and High-Accuracy Switchable Stereo Depth Estimation Method Utilizing Self-Supervised Online Learning Mechanism for MIS
In minimally invasive surgery (MIS), clinicians often rely on 2-D laparoscopic images to assess the size and distances between internal structures. However, this subjective estimation introduces significant uncertainty and can increase surgical risks while reducing efficiency. Modern 3-D laparoscopes offer improved stereoscopic perception and can incorporate stereo depth estimation methods for quantitative analysis. However, existing methods struggle with real-time and high-accuracy demands in diverse surgical scenarios. To address this issue, we propose a novel intraoperative stereo depth estimation framework termed metainitialized online learning (MIOL), aiming to assist surgeons in quantitatively controlling surgical targets during the procedure. This framework features two switchable modes and does not require annotated data. One mode enables rapid depth recovery through surgical videos, providing real-time 3-D reconstruction to help surgeons understand in vivo structures. The other mode achieves high-precision measurements of critical tissues in fixed frames, assisting in surgical decision-making. Our approach employs self-supervised adaptation to train a model specific to each stereo image, eliminating the need for generalization and achieving outstanding accuracy. The framework establishes a lightweight network that converges rapidly under self-supervised losses and incorporates meta-learning pretraining, sparse optical flow guidance, and effective region identification to ensure speed and accuracy. Extensive experiments on two public datasets demonstrate the superiority of our method over existing approaches. Furthermore, we implement an intraoperative measurement system and conduct clinical trials, confirming its practical utility. The code is available at https://github.com/Darcy-vision/MIOL .
- 5 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-13
Real-Time and High-Accuracy Switchable Stereo Depth Estimation Method Utilizing Self-Supervised Online Learning Mechanism for MIS
In minimally invasive surgery (MIS), clinicians often rely on 2-D laparoscopic images to assess the size and distances between internal structures. However, this subjective estimation introduces significant uncertainty and can increase surgical risks while reducing efficiency. Modern 3-D laparoscopes offer improved stereoscopic perception and can incorporate stereo depth estimation methods for quantitative analysis. However, existing methods struggle with real-time and high-accuracy demands in diverse surgical scenarios. To address this issue, we propose a novel intraoperative stereo depth estimation framework termed metainitialized online learning (MIOL), aiming to assist surgeons in quantitatively controlling surgical targets during the procedure. This framework features two switchable modes and does not require annotated data. One mode enables rapid depth recovery through surgical videos, providing real-time 3-D reconstruction to help surgeons understand in vivo structures. The other mode achieves high-precision measurements of critical tissues in fixed frames, assisting in surgical decision-making. Our approach employs self-supervised adaptation to train a model specific to each stereo image, eliminating the need for generalization and achieving outstanding accuracy. The framework establishes a lightweight network that converges rapidly under self-supervised losses and incorporates meta-learning pretraining, sparse optical flow guidance, and effective region identification to ensure speed and accuracy. Extensive experiments on two public datasets demonstrate the superiority of our method over existing approaches. Furthermore, we implement an intraoperative measurement system and conduct clinical trials, confirming its practical utility. The code is available at https://github.com/Darcy-vision/MIOL .