Jingwen Yu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jingwen Yu has a total of 44 co-authors. Here are the top co-authors by number of shared publications:
- Hong Zhang
Chinese Academy of Sciences
4 shared publications - Guangcheng Chen
Southern University of Science and Technology
3 shared publications - Hanjing Ye
Shenzhen Key Laboratory of Robotics and Computer Vision, Southern University of Science and Technology, Shenzhen, China
2 shared publications - Jianhao Jiao
Department of Computer Science, University College London, London, U.K.
2 shared publications - Jiankun Wang
Southern University of Science and Technology
2 shared publications - Zhonghang Liu
SmartMore Corporation
2 shared publications
Latest Publications
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (12) , 14788-14795
Cross-Modality Diffusion for Descriptor Manifold Translation in Visible-Infrared Matching
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (12) , 14788-14795
Cross-Modality Diffusion for Descriptor Manifold Translation in Visible-Infrared Matching
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (10) , 11561-11568
VG-Mapping: Variation-Aware Density Control for Online 3D Gaussian Mapping in Semi-Static Scenes
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (10) , 11561-11568
VG-Mapping: Variation-Aware Density Control for Online 3D Gaussian Mapping in Semi-Static Scenes
- 10 September 2026
- Published by Springer Nature
- p. 491-510
VUOD: A Versatile Unsupervised Outlier Detection Framework for Natural, Industrial, Medical Images and Beyond
- 10 September 2026
- Published by Springer Nature
- p. 491-510
VUOD: A Versatile Unsupervised Outlier Detection Framework for Natural, Industrial, Medical Images and Beyond
- 8 June 2026
- Published by SAGE Publications in The International Journal of Robotics Research
TPT-Bench: A large-scale, long-term and robot-egocentric dataset for benchmarking target person tracking
Tracking a target person from robot-egocentric views is crucial for developing autonomous robots that provide continuous personalized assistance or collaboration in human–robot interaction (HRI) and Embodied AI. However, most existing target person tracking (TPT) benchmarks are limited to controlled laboratory environments with few distractions, clean backgrounds, and short-term occlusions. In this paper, we introduce a large-scale dataset designed for TPT in crowded and unstructured environments, demonstrated through a robot-person following task. The dataset is collected by a human pushing a sensor-equipped cart while following a target person, capturing human-like following behavior and emphasizing long-term tracking challenges, including frequent occlusions and the need for re-identification from numerous pedestrians. It includes multi-modal data streams, including odometry, 3D LiDAR, IMU, panoramic images, and RGB-D images, along with exhaustively annotated 2D bounding boxes of the target person across 48 sequences, both indoors and outdoors. Using this dataset and visual annotations, we perform extensive experiments with existing SOTA TPT methods, offering a thorough analysis of their limitations and suggesting future research directions. Our dataset, code, and video are available at https://medlartea.github.io/tpt-bench/ .
- 8 June 2026
- Published by SAGE Publications in The International Journal of Robotics Research
TPT-Bench: A large-scale, long-term and robot-egocentric dataset for benchmarking target person tracking
Tracking a target person from robot-egocentric views is crucial for developing autonomous robots that provide continuous personalized assistance or collaboration in human–robot interaction (HRI) and Embodied AI. However, most existing target person tracking (TPT) benchmarks are limited to controlled laboratory environments with few distractions, clean backgrounds, and short-term occlusions. In this paper, we introduce a large-scale dataset designed for TPT in crowded and unstructured environments, demonstrated through a robot-person following task. The dataset is collected by a human pushing a sensor-equipped cart while following a target person, capturing human-like following behavior and emphasizing long-term tracking challenges, including frequent occlusions and the need for re-identification from numerous pedestrians. It includes multi-modal data streams, including odometry, 3D LiDAR, IMU, panoramic images, and RGB-D images, along with exhaustively annotated 2D bounding boxes of the target person across 48 sequences, both indoors and outdoors. Using this dataset and visual annotations, we perform extensive experiments with existing SOTA TPT methods, offering a thorough analysis of their limitations and suggesting future research directions. Our dataset, code, and video are available at https://medlartea.github.io/tpt-bench/ .
- 16 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 23 (15455955) , 8476-8489
ROVER: Robust Loop Closure Verification With Trajectory Prior in Repetitive Environments
Loop closure detection is important for simultaneous localization and mapping (SLAM), which associates current observations with historical keyframes, achieving drift correction and global relocalization. However, a falsely detected loop can be fatal, and this is especially difficult in repetitive environments where appearance-based features fail due to the high similarity. Therefore, verifying a loop closure is a critical step to avoid false-positive detections. Existing works in loop closure verification predominantly focus on learning invariant appearance features, neglecting the prior knowledge of the robot’s spatial-temporal motion cue, i.e., trajectory. In this article, we propose ROVER, a loop closure verification method that leverages the historical trajectory as a prior constraint to reject false loops in challenging repetitive environments. For each loop candidate, it is first used to estimate the robot trajectory with pose-graph optimization. This trajectory is then submitted to a scoring scheme that assesses its compliance with the trajectory without the loop, which we refer to as the trajectory prior constraint (TPC), to determine if the loop candidate should be accepted. Benchmark comparisons and real-world experiments demonstrate the effectiveness of the proposed method. Furthermore, we integrate ROVER into state-of-the-art SLAM systems to verify its robustness and efficiency. Our source code and self-collected dataset will be made available online at https://rover-lcv.github.io/ upon publication of this article.
- 16 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 23 (15455955) , 8476-8489
ROVER: Robust Loop Closure Verification With Trajectory Prior in Repetitive Environments
Loop closure detection is important for simultaneous localization and mapping (SLAM), which associates current observations with historical keyframes, achieving drift correction and global relocalization. However, a falsely detected loop can be fatal, and this is especially difficult in repetitive environments where appearance-based features fail due to the high similarity. Therefore, verifying a loop closure is a critical step to avoid false-positive detections. Existing works in loop closure verification predominantly focus on learning invariant appearance features, neglecting the prior knowledge of the robot’s spatial-temporal motion cue, i.e., trajectory. In this article, we propose ROVER, a loop closure verification method that leverages the historical trajectory as a prior constraint to reject false loops in challenging repetitive environments. For each loop candidate, it is first used to estimate the robot trajectory with pose-graph optimization. This trajectory is then submitted to a scoring scheme that assesses its compliance with the trajectory without the loop, which we refer to as the trajectory prior constraint (TPC), to determine if the loop candidate should be accepted. Benchmark comparisons and real-world experiments demonstrate the effectiveness of the proposed method. Furthermore, we integrate ROVER into state-of-the-art SLAM systems to verify its robustness and efficiency. Our source code and self-collected dataset will be made available online at https://rover-lcv.github.io/ upon publication of this article.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (9) , 1-8
TARIC: Memory-Augmented Traversability-Aware Outdoor VLN Under Interrupted Semantic Cues
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (9) , 1-8
TARIC: Memory-Augmented Traversability-Aware Outdoor VLN Under Interrupted Semantic Cues
- 24 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 23 (15455955) , 401-416
Asymptotically Optimal Lifelong Planning With Lazy Edge Evaluation Under Expensive Collision Checks
Robotic systems operating in dynamic and uncertain environments require motion planners that can rapidly adapt to environmental changes while maintaining safety and efficiency. Frequent replanning is inevitable in such scenarios as obstacle configurations evolve and previously feasible trajectories become invalid. However, real-time replanning remains challenging for many applications due to the high computational cost of collision checking and graph maintenance, especially when edge evaluations are expensive or when the environment changes continuously. The paper introduces an asymptotically optimal lifelong sampling-based path planning algorithm that combines the merits of lifelong planning algorithms and lazy search algorithms for rapid replanning in dynamic environments where edge evaluation is expensive. The algorithm maintains an incremental search graph which is reused throughout the entire navigation process. By evaluating only sub-path candidates for the optimal solution, the algorithm saves considerable evaluation time and reduces the overall planning cost. It employs a novel informed rewiring cascade to efficiently repair the search tree when the underlying search graph changes. Theoretical analysis indicates that the proposed algorithm converges to the optimal solution as long as sufficient planning time is given. Planning results on robotic systems with SE(3) and R7 state spaces in challenging environments highlight the superior performance of the proposed algorithm over various state-of-the-art sampling-based planners in both static and dynamic motion planning tasks. The experiment of planning for a Turtlebot 4 operating in a dynamic environment with several moving pedestrians further verifies the feasibility and advantages of the proposed algorithm.
- 24 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automation Science and Engineering
- Vol. 23 (15455955) , 401-416
Asymptotically Optimal Lifelong Planning With Lazy Edge Evaluation Under Expensive Collision Checks
Robotic systems operating in dynamic and uncertain environments require motion planners that can rapidly adapt to environmental changes while maintaining safety and efficiency. Frequent replanning is inevitable in such scenarios as obstacle configurations evolve and previously feasible trajectories become invalid. However, real-time replanning remains challenging for many applications due to the high computational cost of collision checking and graph maintenance, especially when edge evaluations are expensive or when the environment changes continuously. The paper introduces an asymptotically optimal lifelong sampling-based path planning algorithm that combines the merits of lifelong planning algorithms and lazy search algorithms for rapid replanning in dynamic environments where edge evaluation is expensive. The algorithm maintains an incremental search graph which is reused throughout the entire navigation process. By evaluating only sub-path candidates for the optimal solution, the algorithm saves considerable evaluation time and reduces the overall planning cost. It employs a novel informed rewiring cascade to efficiently repair the search tree when the underlying search graph changes. Theoretical analysis indicates that the proposed algorithm converges to the optimal solution as long as sufficient planning time is given. Planning results on robotic systems with SE(3) and R7 state spaces in challenging environments highlight the superior performance of the proposed algorithm over various state-of-the-art sampling-based planners in both static and dynamic motion planning tasks. The experiment of planning for a Turtlebot 4 operating in a dynamic environment with several moving pedestrians further verifies the feasibility and advantages of the proposed algorithm.
- Open Access
- 17 December 2024
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 44 (7) , 1093-1116
FusionPortableV2: A unified multi-sensor dataset for generalized SLAM across diverse platforms and scalable environments
Simultaneous Localization and Mapping (SLAM) has been widely applied in various robotic missions, from rescue operations to autonomous driving. However, the generalization of SLAM algorithms remains a significant challenge, as current datasets often lack scalability in terms of platforms and environments. To address this limitation, we present FusionPortableV2, a multi-sensor SLAM dataset featuring sensor diversity, varied motion patterns, and a wide range of environmental scenarios. Our dataset comprises 27 sequences, spanning over 2.5 hours and collected from four distinct platforms: a handheld suite, a legged robot, an unmanned ground vehicle (UGV), and a vehicle. These sequences cover diverse settings, including buildings, campuses, and urban areas, with a total length of 38.7 km. Additionally, the dataset includes ground truth (GT) trajectories and RGB point cloud maps covering approximately 0.3 km 2 . To validate the utility of our dataset in advancing SLAM research, we assess several state-of-the-art (SOTA) SLAM algorithms. Furthermore, we demonstrate the dataset’s broad application beyond traditional SLAM tasks by investigating its potential for monocular depth estimation. The complete dataset, including sensor data, GT, and calibration details, is accessible at https://fusionportable.github.io/dataset/fusionportable_v2 .
- 17 December 2024
- Published by SAGE Publications in The International Journal of Robotics Research
- Vol. 44 (7) , 1093-1116
Open AccessFusionPortableV2: A unified multi-sensor dataset for generalized SLAM across diverse platforms and scalable environments
Simultaneous Localization and Mapping (SLAM) has been widely applied in various robotic missions, from rescue operations to autonomous driving. However, the generalization of SLAM algorithms remains a significant challenge, as current datasets often lack scalability in terms of platforms and environments. To address this limitation, we present FusionPortableV2, a multi-sensor SLAM dataset featuring sensor diversity, varied motion patterns, and a wide range of environmental scenarios. Our dataset comprises 27 sequences, spanning over 2.5 hours and collected from four distinct platforms: a handheld suite, a legged robot, an unmanned ground vehicle (UGV), and a vehicle. These sequences cover diverse settings, including buildings, campuses, and urban areas, with a total length of 38.7 km. Additionally, the dataset includes ground truth (GT) trajectories and RGB point cloud maps covering approximately 0.3 km 2 . To validate the utility of our dataset in advancing SLAM research, we assess several state-of-the-art (SOTA) SLAM algorithms. Furthermore, we demonstrate the dataset’s broad application beyond traditional SLAM tasks by investigating its potential for monocular depth estimation. The complete dataset, including sensor data, GT, and calibration details, is accessible at https://fusionportable.github.io/dataset/fusionportable_v2 .
- 26 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73 (2) , 2292-2304
Cloud Learning-Based Meets Edge Model-Based: Robots Don't Need to Build All the Submaps Itself
In recent years, significant progress has been made in learning-based VSLAM (Visual Simultaneous Localization and Mapping). Cloud-based VSLAM is a promising solution for meeting the computational demands of learning-based methods in mobile robot applications. However, existing cloud-based VSLAM systems face high transmission demands. To address this issue, we propose a cloud-based VSLAM system, offloading the heavy cost of reconstructing challenging images to the cloud using the learning-based method and leaving the light realtime tracking in the edge using the model-based method. By combining the cloud-edge transmission and a multiple submap VSLAM framework, we introduce a rumination-inspired mechanism for asynchronous and distributed submap building. The submap-based framework and proposed down-sampling method help reduce transmission frequency and data volume. We present experimental results that demonstrate the robustness and precision of our cloud-based multiple submap VSLAM system. We also evaluate the runtime performance of communication and computation on a real robot platform, which suggests that the multiple submap VSLAM framework can effectively release computation load while satisfying both robustness and realtime requirements.
- 26 September 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 73 (2) , 2292-2304
Cloud Learning-Based Meets Edge Model-Based: Robots Don't Need to Build All the Submaps Itself
In recent years, significant progress has been made in learning-based VSLAM (Visual Simultaneous Localization and Mapping). Cloud-based VSLAM is a promising solution for meeting the computational demands of learning-based methods in mobile robot applications. However, existing cloud-based VSLAM systems face high transmission demands. To address this issue, we propose a cloud-based VSLAM system, offloading the heavy cost of reconstructing challenging images to the cloud using the learning-based method and leaving the light realtime tracking in the edge using the model-based method. By combining the cloud-edge transmission and a multiple submap VSLAM framework, we introduce a rumination-inspired mechanism for asynchronous and distributed submap building. The submap-based framework and proposed down-sampling method help reduce transmission frequency and data volume. We present experimental results that demonstrate the robustness and precision of our cloud-based multiple submap VSLAM system. We also evaluate the runtime performance of communication and computation on a real robot platform, which suggests that the multiple submap VSLAM framework can effectively release computation load while satisfying both robustness and realtime requirements.