Linhu Wei
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Linhu Wei has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Aiguo Song
Southeast University
5 shared publications - Huiran Hu
School of Instrument Science & Engineering, Southeast University, China
3 shared publications - Jingchao Mao
School of Instrument Science and Engineering, Southeast University, China
3 shared publications - Baoguo Xu
Southeast University
1 shared publication - Huailiang Ma
Southeast University
1 shared publication - Jiajin Wang
State Key Laboratory of Digital Medical Engineering, School of Instrument Science and Engineering, Southeast University, Nanjing, China
1 shared publication
Latest Publications
- 2 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-20
Point Cloud Perception in Human-to-Robot Handover Based on Semantic Segmentation and Shape Completion
Effective object handovers from humans-to-robots (H2R) in multi-scenario environments are a key challenge in human-robot collaboration. Numerous studies have explored various perception strategies. However, robust handover in complex settings remains challenging due to frequent lighting interference to RGB-based semantic segmentation, or inherent blindspot issue for limited perceptual views. Under these difficulties, this paper proposes a new perception framework for human-robot handover under eye-on-scene setting. Firstly, this framework uses a single-pixel RGB tracking for initial localization, a point cloud segmentation with regulated extraction, and a learning-based completion for camera blind spots. It works reliably encountering RGB overexposure and shows clear improvement in human handover postures at blindspot. Then, the method introduces a KNN-based loss to improve RandLA-Net segmentation, and an object feature classification strategy to enhance ODGNet-based completion. These boost recognition efficiency and generalization in real handover tasks. Finally, in handover experiments on 15 novel objects, our system achieved an overall success rate of 88.5%. In the quantitative evaluation involving four objects and five representative human handover postures, our method demonstrates improvements over the No Completion baseline in terms of both the overall success rate and the number of predicted valid grasp poses. When encountering RGB overexposure, our method exhibits a more significant advantage in perceptual robustness and grasping effectiveness compared to traditional RGB-based segmentation methods.
- 2 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-20
Point Cloud Perception in Human-to-Robot Handover Based on Semantic Segmentation and Shape Completion
Effective object handovers from humans-to-robots (H2R) in multi-scenario environments are a key challenge in human-robot collaboration. Numerous studies have explored various perception strategies. However, robust handover in complex settings remains challenging due to frequent lighting interference to RGB-based semantic segmentation, or inherent blindspot issue for limited perceptual views. Under these difficulties, this paper proposes a new perception framework for human-robot handover under eye-on-scene setting. Firstly, this framework uses a single-pixel RGB tracking for initial localization, a point cloud segmentation with regulated extraction, and a learning-based completion for camera blind spots. It works reliably encountering RGB overexposure and shows clear improvement in human handover postures at blindspot. Then, the method introduces a KNN-based loss to improve RandLA-Net segmentation, and an object feature classification strategy to enhance ODGNet-based completion. These boost recognition efficiency and generalization in real handover tasks. Finally, in handover experiments on 15 novel objects, our system achieved an overall success rate of 88.5%. In the quantitative evaluation involving four objects and five representative human handover postures, our method demonstrates improvements over the No Completion baseline in terms of both the overall success rate and the number of predicted valid grasp poses. When encountering RGB overexposure, our method exhibits a more significant advantage in perceptual robustness and grasping effectiveness compared to traditional RGB-based segmentation methods.
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (6) , 6935-6942
AutoTrialGen: Automated Data Generation From Few Human Demonstrations via Trajectory Annotation and Simulation Trials
While imitation learning is a powerful paradigm for teaching robots complex manipulation skills, its effectiveness is often bottlenecked by the need for large-scale, human-collected datasets. This paper presents AutoTrialGen, an automated framework designed to generate large and diverse datasets of successful simulated demonstrations through trial-and-error, from only a few human demonstrations. Initially, AutoTrialGen employs a vision-language model to automatically decompose unprocessed human demonstrations into a library of object-centric and reusable skill primitives. Within a simulated environment, it then intelligently composes these primitives to solve new task instances, employing a novel weighted manipulability selection mechanism to ensure the quality and efficiency of the generated trajectories. Policies trained using the proposed sim-augmented data demonstrate improved performance and enhanced data efficiency across six diverse and challenging real-world manipulation tasks, compared with the baseline method.
- 30 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (6) , 6935-6942
AutoTrialGen: Automated Data Generation From Few Human Demonstrations via Trajectory Annotation and Simulation Trials
While imitation learning is a powerful paradigm for teaching robots complex manipulation skills, its effectiveness is often bottlenecked by the need for large-scale, human-collected datasets. This paper presents AutoTrialGen, an automated framework designed to generate large and diverse datasets of successful simulated demonstrations through trial-and-error, from only a few human demonstrations. Initially, AutoTrialGen employs a vision-language model to automatically decompose unprocessed human demonstrations into a library of object-centric and reusable skill primitives. Within a simulated environment, it then intelligently composes these primitives to solve new task instances, employing a novel weighted manipulability selection mechanism to ensure the quality and efficiency of the generated trajectories. Policies trained using the proposed sim-augmented data demonstrate improved performance and enhanced data efficiency across six diverse and challenging real-world manipulation tasks, compared with the baseline method.
- Open Access
- 27 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 77-82
Development of a Virtual Reality-Based Teleoperated Welding Robot System for Enhanced Safety and Efficiency
This paper presents a teleoperated welding robot system developed utilizing virtual reality technology, designed to provide a safe and convenient platform for remote operations to assist welding personnel. The system allows operators to remotely plan and control welding tasks within a high-risk environment through 3D reconstructed welding scenes. It features seamless switching between two operational modes: Path Planning Mode and Teleoperated Welding Mode. In both modes, operators are able to fully observe and control the welding robot and its end effector from a third-person perspective across the virtual reconstructed environment, thereby achieving precise and efficient welding operations. This teleoperation system not only enhances the flexibility of welding operations but also significantly improves job safety.
- 27 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 77-82
Open AccessDevelopment of a Virtual Reality-Based Teleoperated Welding Robot System for Enhanced Safety and Efficiency
This paper presents a teleoperated welding robot system developed utilizing virtual reality technology, designed to provide a safe and convenient platform for remote operations to assist welding personnel. The system allows operators to remotely plan and control welding tasks within a high-risk environment through 3D reconstructed welding scenes. It features seamless switching between two operational modes: Path Planning Mode and Teleoperated Welding Mode. In both modes, operators are able to fully observe and control the welding robot and its end effector from a third-person perspective across the virtual reconstructed environment, thereby achieving precise and efficient welding operations. This teleoperation system not only enhances the flexibility of welding operations but also significantly improves job safety.
- Open Access
- 27 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 96-101
Design Of Intelligent Mobile Robot System Based On Remote Operation
In recent years, with the development of artificial intelligence, mobile robots are gradually moving towards automation and intelligence. In order to address the shortcomings of traditional intelligent mobile robots, such as single task forms and limited application environments, a smart mobile robot system with remote operation capability has been designed to enhance the adaptability of mobile robots to complex environments and tasks. The system consists of three functional modules: autonomous navigation module, remote operation module, and VR-assisted control module. A set of autonomous navigation systems has been developed to achieve navigation and obstacle avoidance in complex environments. Through the remote operation system, remote control of intelligent mobile robots is realized, enabling autonomous navigation, obstacle avoidance, and providing a unified communication topic interface to achieve the generality and portability of the remote operation system. Meanwhile, VR head-mounted display devices can assist users in performing complex tasks with intelligent mobile robots, improving the quality and efficiency of task completion.
- 27 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 96-101
Open AccessDesign Of Intelligent Mobile Robot System Based On Remote Operation
In recent years, with the development of artificial intelligence, mobile robots are gradually moving towards automation and intelligence. In order to address the shortcomings of traditional intelligent mobile robots, such as single task forms and limited application environments, a smart mobile robot system with remote operation capability has been designed to enhance the adaptability of mobile robots to complex environments and tasks. The system consists of three functional modules: autonomous navigation module, remote operation module, and VR-assisted control module. A set of autonomous navigation systems has been developed to achieve navigation and obstacle avoidance in complex environments. Through the remote operation system, remote control of intelligent mobile robots is realized, enabling autonomous navigation, obstacle avoidance, and providing a unified communication topic interface to achieve the generality and portability of the remote operation system. Meanwhile, VR head-mounted display devices can assist users in performing complex tasks with intelligent mobile robots, improving the quality and efficiency of task completion.
- Open Access
- 27 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 131-137
Mixed Reality-Augmented Remote Welding System with Virtual Fixtures and Autonomous Agents
In modern industrial production, although welding robots surpass humans in stability and safety, they have limitations in dealing with the unpredictability of actual production, especially in unstructured environments. To combine the adaptability of human welders with the precision of robots, we design an intuitive human-machine interactive remote welding system. In this system, we develop a mixed reality interface that allows operators to real-time reconstruct and update the 3D model of unknown workpieces, and integrate virtual fixtures to provide real-time welding control. At the same time, to reduce the cognitive burden on operators, we introduce autonomous agents to automatically maintain welding angles and heights. Experiments conducted on the remote robot welding platform show that compared to traditional direct teleoperation, we effectively reduce the operational load, while reducing trajectory errors by 67.59% and shortening task time by 51%.
- 27 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 131-137
Open AccessMixed Reality-Augmented Remote Welding System with Virtual Fixtures and Autonomous Agents
In modern industrial production, although welding robots surpass humans in stability and safety, they have limitations in dealing with the unpredictability of actual production, especially in unstructured environments. To combine the adaptability of human welders with the precision of robots, we design an intuitive human-machine interactive remote welding system. In this system, we develop a mixed reality interface that allows operators to real-time reconstruct and update the 3D model of unknown workpieces, and integrate virtual fixtures to provide real-time welding control. At the same time, to reduce the cognitive burden on operators, we introduce autonomous agents to automatically maintain welding angles and heights. Experiments conducted on the remote robot welding platform show that compared to traditional direct teleoperation, we effectively reduce the operational load, while reducing trajectory errors by 67.59% and shortening task time by 51%.