Haeseong Lee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Haeseong Lee has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Jaeheung Park
Department of Intelligence and Information, Seoul National University, Seoul, Republic of Korea
9 shared publications - Keunwoo Jang
Graduate School of Convergence Science and Technology, Seoul National University, Seoul, Republic of Korea
3 shared publications - Seungyeon Kim
Graduate School of Convergence Science and Technology, Seoul National University, Seoul, Republic of Korea
3 shared publications - Suhan Park
School of Robotics, Kwangwoon University, Seoul 01897, Republic of Korea
3 shared publications - Eunho Sung
Department of Intelligence and Information, Graduate School of Convergence Science and Technology, Seoul National University, Seoul, Republic of Korea
2 shared publications Chaiyong Mun
Department of Intelligence and Information, Graduate School of Convergence Science and Technology, Seoul National University, Seoul, Republic of Korea
1 shared publication
Latest Publications
- Open Access
- 17 November 2026
- Published by Springer Nature in International Journal of Control, Automation, and Systems
Active Vision for Visuomotor Policy Generalization Under Positional Variations
In recent years, visuomotor policy learning has gained increased attention in the robotics domain. Ideally, such policies enable robots to generate actions directly from observations with high efficiency and accuracy. In reality, performance is significantly degraded when the robot encounters previously unseen environments (e.g., a novel position of an object to grasp, disturbances in the camera viewpoint). Several studies have attempted to deal with this problem by training the model on large-scale datasets, but constructing such datasets raises another challenge of data efficiency. To address these limitations, this work presents a data-efficient visuomotor policy framework via active vision. To be specific, the proposed framework employs an object detection algorithm and expresses the robot state and actions in terms of the camera coordinate frame. Provided that the camera viewpoint is properly adjusted by active vision, the visuomotor policy can manipulate objects at any position, even at positions not included in the training data. The experiments performed in both simulation and real-world environments with a humanoid robot and a dual manipulator system highlight that the proposed framework outperforms the state-of-the-art approaches significantly by enhancing the success rate for the pick-and-place task at unseen object positions from 45 to 91%.
- 17 November 2026
- Published by Springer Nature in International Journal of Control, Automation, and Systems
Open AccessActive Vision for Visuomotor Policy Generalization Under Positional Variations
In recent years, visuomotor policy learning has gained increased attention in the robotics domain. Ideally, such policies enable robots to generate actions directly from observations with high efficiency and accuracy. In reality, performance is significantly degraded when the robot encounters previously unseen environments (e.g., a novel position of an object to grasp, disturbances in the camera viewpoint). Several studies have attempted to deal with this problem by training the model on large-scale datasets, but constructing such datasets raises another challenge of data efficiency. To address these limitations, this work presents a data-efficient visuomotor policy framework via active vision. To be specific, the proposed framework employs an object detection algorithm and expresses the robot state and actions in terms of the camera coordinate frame. Provided that the camera viewpoint is properly adjusted by active vision, the visuomotor policy can manipulate objects at any position, even at positions not included in the training data. The experiments performed in both simulation and real-world environments with a humanoid robot and a dual manipulator system highlight that the proposed framework outperforms the state-of-the-art approaches significantly by enhancing the success rate for the pick-and-place task at unseen object positions from 45 to 91%.
- Open Access
- 4 November 2026
- Published by Springer Nature in International Journal of Control, Automation, and Systems
Software-Defined Context-Aware Virtual Fixture for Teleoperation via Vision-Language Model
Precise teleoperation in unstructured environments remains challenging, because human errors and system-inherent limitations such as control jittering can significantly degrade task success and operational efficiency. To address these limitations, this paper proposes a software-defined Context-Aware Virtual Fixture (CAVF) system that leverages the open-world understanding of a Vision-Language Model (VLM). The proposed system automatically generates motion guidance for the robot end-effector (EEF) to assist teleoperation, preserving critical task-relevant motion while suppressing unnecessary translation and rotation components. This framework enables the autonomous generation of context-aware, haptic-free assistance without relying on handcrafted rules or physical force-feedback devices. Quantitative experiments in both simulation and real-world environments demonstrate that the proposed system significantly improves success rates, reduces completion times, and enhances trajectory efficiency compared to conventional teleoperation.
- 4 November 2026
- Published by Springer Nature in International Journal of Control, Automation, and Systems
Open AccessSoftware-Defined Context-Aware Virtual Fixture for Teleoperation via Vision-Language Model
Precise teleoperation in unstructured environments remains challenging, because human errors and system-inherent limitations such as control jittering can significantly degrade task success and operational efficiency. To address these limitations, this paper proposes a software-defined Context-Aware Virtual Fixture (CAVF) system that leverages the open-world understanding of a Vision-Language Model (VLM). The proposed system automatically generates motion guidance for the robot end-effector (EEF) to assist teleoperation, preserving critical task-relevant motion while suppressing unnecessary translation and rotation components. This framework enables the autonomous generation of context-aware, haptic-free assistance without relying on handcrafted rules or physical force-feedback devices. Quantitative experiments in both simulation and real-world environments demonstrate that the proposed system significantly improves success rates, reduces completion times, and enhances trajectory efficiency compared to conventional teleoperation.
- 12 June 2026
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 204
Bi-spiral: Search strategy for peg-in-hole assembly using dual manipulator system
- 12 June 2026
- Published by Elsevier in Robotics and Autonomous Systems
- Vol. 204
Bi-spiral: Search strategy for peg-in-hole assembly using dual manipulator system
- Open Access
- 13 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 57409-57426
SNU-Avatar Haptic Arm: A Hybrid QDD/DD Operator Interface for Teleoperation
Despite rapid advancements in robotics, teleoperation remains indispensable for precise tasks that have not yet been fully automated. To this end, this paper presents the design, analysis, and experimental validation of the SNU-Avatar Haptic Arm, a high-performance haptic interface developed for immersive teleoperation. To balance payload capacity and backdrivability, the system employs a hybrid actuation architecture combining quasi-direct drive and direct drive mechanisms. This configuration reduces mechanical impedance and friction while ensuring sufficient force output for high-fidelity interaction. The 6-degrees-of-freedom kinematic structure is designed to encompass the natural workspace of the human upper limb, and finite element analysis confirms that the structural natural frequency is sufficiently high to secure a stable control bandwidth. A comprehensive evaluation validates the system’s performance, focusing on force rendering accuracy. The practical utility of the proposed interface was demonstrated in the ANA Avatar XPRIZE Finals through integration with a humanoid robot. The system successfully performed complex missions, such as heavy object manipulation, under constraints including communication latency. The results confirm that the proposed design methodology effectively bridges the gap between remote presence and physical reality, ensuring robust manipulation in extreme environments.
- 13 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 57409-57426
Open AccessSNU-Avatar Haptic Arm: A Hybrid QDD/DD Operator Interface for Teleoperation
Despite rapid advancements in robotics, teleoperation remains indispensable for precise tasks that have not yet been fully automated. To this end, this paper presents the design, analysis, and experimental validation of the SNU-Avatar Haptic Arm, a high-performance haptic interface developed for immersive teleoperation. To balance payload capacity and backdrivability, the system employs a hybrid actuation architecture combining quasi-direct drive and direct drive mechanisms. This configuration reduces mechanical impedance and friction while ensuring sufficient force output for high-fidelity interaction. The 6-degrees-of-freedom kinematic structure is designed to encompass the natural workspace of the human upper limb, and finite element analysis confirms that the structural natural frequency is sufficiently high to secure a stable control bandwidth. A comprehensive evaluation validates the system’s performance, focusing on force rendering accuracy. The practical utility of the proposed interface was demonstrated in the ANA Avatar XPRIZE Finals through integration with a humanoid robot. The system successfully performed complex missions, such as heavy object manipulation, under constraints including communication latency. The results confirm that the proposed design methodology effectively bridges the gap between remote presence and physical reality, ensuring robust manipulation in extreme environments.
- 31 August 2024
- Published by The Korea Robotics Society in Journal of Korea Robotics Society
- Vol. 19 (3) , 236-243
Development of Collaborative Dual Manipulator System for Packaging Industrial Coils
- 31 August 2024
- Published by The Korea Robotics Society in Journal of Korea Robotics Society
- Vol. 19 (3) , 236-243
Development of Collaborative Dual Manipulator System for Packaging Industrial Coils
- 25 April 2024
- Published by Springer Nature
- p. 521-535
Walking State Estimation for Biped Robot Using Foot Contact Information
- 25 April 2024
- Published by Springer Nature
- p. 521-535
Walking State Estimation for Biped Robot Using Foot Contact Information
- 23 May 2022
- Published by Springer Nature in Intelligent Service Robotics
- Vol. 15 (4) , 441-457
Robotic furniture assembly: task abstraction, motion planning, and control
- 23 May 2022
- Published by Springer Nature in Intelligent Service Robotics
- Vol. 15 (4) , 441-457
Robotic furniture assembly: task abstraction, motion planning, and control
- 31 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 7 (2) , 3349-3356
Contact State Estimation for Peg-in-Hole Assembly Using Gaussian Mixture Model
Recently, the robotic assembly has been expanded into an unstructured environment. This environment includes uncertainties that may cause unexpected situations such as a failure of the assembly. Such problems can be prevented or monitored by a robust contact state (CS) estimation method. In that sense, the paper suggests a CS estimation method that contains a torque indicator, a position/velocity indicator, and a CS discriminator. Using joint torque of manipulators and position/velocity of the end-effector, a Gaussian Mixture Model (GMM) builds each indicator by reflecting on two properties of measured data, i.e., non-stationary behavior and correlation among the data. The indicators play a role to indicate the corresponding sensor state. The discriminator is defined by rules which combine the results of the indicators, allowing a robust CS estimation to be achieved. In this respect, the proposed method has a distinct advantage over existing distance-based clustering methods which ignore probabilistic properties or correlation among measured data. The performance of the estimation is demonstrated through experiments with torque-controlled manipulators and commercial prefabricated furniture.
- 31 January 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 7 (2) , 3349-3356
Contact State Estimation for Peg-in-Hole Assembly Using Gaussian Mixture Model
Recently, the robotic assembly has been expanded into an unstructured environment. This environment includes uncertainties that may cause unexpected situations such as a failure of the assembly. Such problems can be prevented or monitored by a robust contact state (CS) estimation method. In that sense, the paper suggests a CS estimation method that contains a torque indicator, a position/velocity indicator, and a CS discriminator. Using joint torque of manipulators and position/velocity of the end-effector, a Gaussian Mixture Model (GMM) builds each indicator by reflecting on two properties of measured data, i.e., non-stationary behavior and correlation among the data. The indicators play a role to indicate the corresponding sensor state. The discriminator is defined by rules which combine the results of the indicators, allowing a robust CS estimation to be achieved. In this respect, the proposed method has a distinct advantage over existing distance-based clustering methods which ignore probabilistic properties or correlation among measured data. The performance of the estimation is demonstrated through experiments with torque-controlled manipulators and commercial prefabricated furniture.
- 4 September 2021
- Published by Springer Nature in Intelligent Service Robotics
- Vol. 14 (4) , 597-609
Search trajectory with twisting motion for dual peg-in-hole assembly
- 4 September 2021
- Published by Springer Nature in Intelligent Service Robotics
- Vol. 14 (4) , 597-609
Search trajectory with twisting motion for dual peg-in-hole assembly