Publications

Publications in the Last Five Years

2026

Preprints and Manuscripts

QUAD: Q-Gradient Uncertainty-Aware Guidance for Diffusion Policies in Offline Reinforcement Learning

Hongtu Zhou, Qian Chen, Junqiao Zhao*, Haoqi Zhao, Changhao Yin, Qingjun Wang, Hang Yu, Ziqiao Wang, Jun Wang, Chen Ye and Guang Chen

arXiv, 2026.

BibTeX
@article{Zhou2026,
	author = {Hongtu Zhou and Qian Chen and Junqiao Zhao* and Haoqi Zhao and Changhao Yin and Qingjun Wang and Hang Yu and Ziqiao Wang and Jun Wang and Chen Ye and Guang Chen},
	title = {QUAD: Q-Gradient Uncertainty-Aware Guidance for Diffusion Policies in Offline Reinforcement Learning},
	journal = {arXiv},
	year = {2026}
}

GeoUniPR: A Geometry-Consistent Unified Framework for Cross-Modal Place Recognition

Wonbong Kim, Jiatong Xiao, Rui Li, Xufei Wang, Qiwen Gu, Junqiao Zhao*, Chen Ye and Guang Chen

arXiv, 2026.

BibTeX
@article{Kim2026,
	author = {Wonbong Kim and Jiatong Xiao and Rui Li and Xufei Wang and Qiwen Gu and Junqiao Zhao* and Chen Ye and Guang Chen},
	title = {GeoUniPR: A Geometry-Consistent Unified Framework for Cross-Modal Place Recognition},
	journal = {arXiv},
	year = {2026}
}

Learning Geometrically-Grounded Amodal 3D Representations for View-Generalizable Robotic Manipulation

Di Zhang, Weicheng Duan, Dasen Gu, Hongye Lu, Hai Zhang, Hang Yu, Guang Chen, Chuan Wen and Junqiao Zhao*

arXiv, 2026.

Link
BibTeX
@article{Di2026,
	author = {Di Zhang and Weicheng Duan and Dasen Gu and Hongye Lu and Hai Zhang and Hang Yu and Guang Chen and Chuan Wen and Junqiao Zhao*},
	title = {Learning Geometrically-Grounded Amodal 3D Representations for View-Generalizable Robotic Manipulation},
	journal = {arXiv},
	year = {2026},
	url = {https://arxiv.org/abs/2601.22988}
}

R2M-Bench: Evaluating Revisit Memory via Relative Consistency in Interactive Video World Models

Qiwen Gu, Bingjie Gao, Rui Chen, Geng Li, Jifan Li, Qishuai Wen, Li Niu, Jing Tang, Xiangxiang Chu and Junqiao Zhao*

arXiv, 2026.

Link
BibTeX
@article{R2M-Bench2026,
	author = {Qiwen Gu and Bingjie Gao and Rui Chen and Geng Li and Jifan Li and Qishuai Wen and Li Niu and Jing Tang and Xiangxiang Chu and Junqiao Zhao*},
	title = {R2M-Bench: Evaluating Revisit Memory via Relative Consistency in Interactive Video World Models},
	journal = {arXiv},
	year = {2026},
	url = {https://arxiv.org/abs/2608.27328},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

SC-JSRL: Training-Safe Policy Improvement via Certified Handoff

Jialin Liu, Qian Chen and Junqiao Zhao*

Unpublished manuscript, 2026.

BibTeX
@article{SC-JSRL2026,
	author = {Jialin Liu and Qian Chen and Junqiao Zhao*},
	title = {SC-JSRL: Training-Safe Policy Improvement via Certified Handoff},
	journal = {Unpublished manuscript},
	year = {2026},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

ACSAC: Adaptive Chunk Size Actor-Critic with Causal Transformer Q-Network

Qian Chen, Junqiao Zhao*, Hongtu Zhou, Hang Yu, Yanping Zhao, Chen Ye and Guang Chen

arXiv, 2026.

Link
BibTeX
@article{ACSAC2026,
	author = {Qian Chen and Junqiao Zhao* and Hongtu Zhou and Hang Yu and Yanping Zhao and Chen Ye and Guang Chen},
	title = {ACSAC: Adaptive Chunk Size Actor-Critic with Causal Transformer Q-Network},
	journal = {arXiv},
	year = {2026},
	url = {https://arxiv.org/abs/2605.11009},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

GWM-VLA: Geometry-Aware Latent World Modeling for Vision-Language-Action Learning

Yanping Zhao, Hang Yu, Yiwei Wang, Chen Ye, Siyu Tian, Di Zhang, Qingjun Wang, Qian Chen, Junqiao Zhao* and Guang Chen

arXiv, 2026.

Link
BibTeX
@article{Yanping2026,
	author = {Yanping Zhao and Hang Yu and Yiwei Wang and Chen Ye and Siyu Tian and Di Zhang and Qingjun Wang and Qian Chen and Junqiao Zhao* and Guang Chen},
	title = {GWM-VLA: Geometry-Aware Latent World Modeling for Vision-Language-Action Learning},
	journal = {arXiv},
	year = {2026},
	url = {https://arxiv.org/abs/2608.07619},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

UG-Diff: Improving Neural Rendering in Distractor-Corrupted Scenes via Uncertainty-Guided Diffusion

Wenjie Mu, Zhan Li, Jingxiao Han, Tong Wu, Ziniu Liu, Chuanzhou Su, Xuanyi Shen, Fan Lu, Junqiao Zhao*, Tiantian Feng, Chen Ye and Guang Chen

Unpublished manuscript, 2026.

BibTeX
@article{UGDiff2026,
	author = {Wenjie Mu and Zhan Li and Jingxiao Han and Tong Wu and Ziniu Liu and Chuanzhou Su and Xuanyi Shen and Fan Lu and Junqiao Zhao* and Tiantian Feng and Chen Ye and Guang Chen},
	title = {UG-Diff: Improving Neural Rendering in Distractor-Corrupted Scenes via Uncertainty-Guided Diffusion},
	journal = {Unpublished manuscript},
	year = {2026},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

PersistSplat: Streaming Feed-Forward Reconstruction with Persistent Gaussian Memory

Ziniu Liu, Wenjie Mu, Jingxiao Han, Zhan Li, Tong Wu, Chuanzhou Su, Xuanyi Shen, Fan Lu, Junqiao Zhao*, Tiantian Feng, Chen Ye and Guang Chen

Unpublished manuscript, 2026.

BibTeX
@article{Ziniu2026,
	author = {Ziniu Liu and Wenjie Mu and Jingxiao Han and Zhan Li and Tong Wu and Chuanzhou Su and Xuanyi Shen and Fan Lu and Junqiao Zhao* and Tiantian Feng and Chen Ye and Guang Chen},
	title = {PersistSplat: Streaming Feed-Forward Reconstruction with Persistent Gaussian Memory},
	journal = {Unpublished manuscript},
	year = {2026},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

HIERAMEM: Evidence-Anchored Hierarchical Memory for Self-Evolving LLM Agents

Xiying Chen, Junqiao Zhao*, Zekai Zhu, Yanping Zhao, Chen Ye, Lianghua He and Guang Chen

Unpublished manuscript, 2026.

BibTeX
@article{Xiying2026,
	author = {Xiying Chen and Junqiao Zhao* and Zekai Zhu and Yanping Zhao and Chen Ye and Lianghua He and Guang Chen},
	title = {HIERAMEM: Evidence-Anchored Hierarchical Memory for Self-Evolving LLM Agents},
	journal = {Unpublished manuscript},
	year = {2026},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

Conference Papers

Beyond Penalization: Diffusion-based Out-of-Distribution Detection and Selective Regularization in Offline Reinforcement Learning

Qingjun Wang, Hongtu Zhou, Hang Yu, Junqiao Zhao*, Yanping Zhao, Chen Ye, Ziqiao Wang and Guang Chen

International Conference on Learning Representations (ICLR 2026), Rio de Janeiro, Brazil, 2026.

TL;DR: Uses diffusion models to detect out-of-distribution actions and selectively regularize them while preserving beneficial actions.

Link
BibTeX
@inproceedings{Wang2026BeyondPenalization,
	author = {Wang, Qingjun and Zhou, Hongtu and Yu, Hang and Zhao*, Junqiao and Zhao, Yanping and Ye, Chen and Wang, Ziqiao and Chen, Guang},
	title = {Beyond Penalization: Diffusion-based Out-of-Distribution Detection and Selective Regularization in Offline Reinforcement Learning},
	booktitle = {International Conference on Learning Representations (ICLR 2026)},
	year = {2026},
	address = {Rio de Janeiro, Brazil},
	url = {https://proceedings.iclr.cc/paper_files/paper/2026/hash/3a2dbdf0aca2006f4751abf0804620c0-Abstract-Conference.html},
	contribution_en = {Uses diffusion models to detect out-of-distribution actions and selectively regularize them while preserving beneficial actions.},
	contribution_zh = {提出基于扩散模型的离线强化学习方法，用于检测分布外动作并有选择地正则化，以更好保留有益动作。}
}

Potentially Optimal Joint Actions Recognition for Cooperative Multi-Agent Reinforcement Learning

Chang Huang, Shatong Zhu, Junqiao Zhao*, Hongtu Zhou, Hai Zhang, Di Zhang, Chen Ye, Ziqiao Wang and Guang Chen

International Conference on Learning Representations (ICLR 2026), Rio de Janeiro, Brazil, 2026.

TL;DR: Identifies potentially optimal joint actions and adaptively adjusts training weights to improve convergence and policy recovery in cooperative multi-agent reinforcement learning.

Link
BibTeX
@inproceedings{Huang2026PotentiallyOptimalJointActions,
	author = {Huang, Chang and Zhu, Shatong and Zhao*, Junqiao and Zhou, Hongtu and Zhang, Hai and Zhang, Di and Ye, Chen and Wang, Ziqiao and Chen, Guang},
	title = {Potentially Optimal Joint Actions Recognition for Cooperative Multi-Agent Reinforcement Learning},
	booktitle = {International Conference on Learning Representations (ICLR 2026)},
	year = {2026},
	address = {Rio de Janeiro, Brazil},
	url = {https://openreview.net/pdf?id=YQ1muQBDV4},
	contribution_en = {Identifies potentially optimal joint actions and adaptively adjusts training weights to improve convergence and policy recovery in cooperative multi-agent reinforcement learning.},
	contribution_zh = {识别可能最优的联合动作，并自适应调整训练权重，以改善协作多智能体强化学习的收敛和策略恢复。}
}

Signal Structure-Aware Gaussian Splatting for Large-Scale Scene Reconstruction

Weiyi Xue, Fan Lu, Chi Zhang, Tianhang Wang, Sanqing Qu, Zehan Zheng, Boyuan Zheng, Junqiao Zhao and Guang Chen

International Conference on Learning Representations (ICLR 2026), Rio de Janeiro, Brazil, 2026.

TL;DR: Introduces a signal-structure-aware Gaussian Splatting framework that improves geometric consistency and reconstruction efficiency for large-scale scenes.

Link
BibTeX
@inproceedings{Xue2026SignalStructureAwareGaussianSplatting,
	author = {Xue, Weiyi and Lu, Fan and Zhang, Chi and Wang, Tianhang and Qu, Sanqing and Zheng, Zehan and Zheng, Boyuan and Zhao, Junqiao and Chen, Guang},
	title = {Signal Structure-Aware Gaussian Splatting for Large-Scale Scene Reconstruction},
	booktitle = {International Conference on Learning Representations (ICLR 2026)},
	year = {2026},
	address = {Rio de Janeiro, Brazil},
	url = {https://proceedings.iclr.cc/paper_files/paper/2026/hash/f67449c7ab72f441d3a713b046c6818c-Abstract-Conference.html},
	contribution_en = {Introduces a signal-structure-aware Gaussian Splatting framework that improves geometric consistency and reconstruction efficiency for large-scale scenes.},
	contribution_zh = {提出信号结构感知的高斯泼溅重建框架，在大规模场景中提升几何一致性与重建效率。}
}

PF-SGS: Pose-Free Streaming 3D Gaussian Splatting for Large-Scale Scene Reconstruction

Wenjie Mu, Ziniu Liu, Tong Wu, Zhan Li, Chuanzhou Su, Xuanyi Shen, Fan Lu, Junqiao Zhao, Tiantian Feng, Chen Ye and Guang Chen

The Fortieth Annual Conference on Neural Information Processing Systems (NeurIPS 2026), 2026.

TL;DR: Proposes a feedforward 3D Gaussian Splatting reconstruction method based on streaming data, introducing global state and gating mechanisms to improve geometric consistency and reconstruction efficiency in long-sequence scenes.

BibTeX
@inproceedings{zotero-item-60448,
	author = {Mu, Wenjie and Liu, Ziniu and Wu, Tong and Li, Zhan and Su, Chuanzhou and Shen, Xuanyi and Lu, Fan and Zhao, Junqiao and Feng, Tiantian and Ye, Chen and Chen, Guang},
	title = {{{PF-SGS}}: {{Pose-Free Streaming 3D Gaussian Splatting}} for {{Large-Scale Scene Reconstruction}}},
	booktitle = {The {{Fortieth Annual Conference}} on {{Neural Information Processing Systems}} ({{NeurIPS}} 2026)},
	year = 2026,
	contribution_en = {Proposes a feedforward 3D Gaussian Splatting reconstruction method based on streaming data, introducing global state and gating mechanisms to improve geometric consistency and reconstruction efficiency in long-sequence scenes.},
	contribution_zh = {提出基于无位姿流式数据的前馈3DGS重建方法，引入全局状态以及门控机制，在长序列场景中提升位姿估计以及几何重建的一致性。}
}

A Unified Framework for Image-to-3D Part Generation via Variable Granularity

Qiwen Gu, Xin Zhou, Xiwu Chen, Zhongyang Zhu, Zhuo Zhang and Junqiao Zhao*

The Fortieth Annual Conference on Neural Information Processing Systems (NeurIPS 2026), 2026.

TL;DR: Proposes a unified approach for generating 3D parts from point-to-mask geometric cues of varying granularity, enhancing cross-modal generation performance while providing flexibility through deep cross-modal fusion during training.

BibTeX
@inproceedings{muPFSGSPoseFreeStreaming2026,
	author = {Gu, Qiwen and Zhou, Xin and Chen, Xiwu and Zhu, Zhongyang and Zhang, Zhuo and Zhao*, Junqiao},
	title = {A {{Unified Framework}} for {{Image}}-to-{{3D Part Generation}} via {{Variable Granularity}}},
	booktitle = {The {{Fortieth Annual Conference}} on {{Neural Information Processing Systems}} ({{NeurIPS}} 2026)},
	year = 2026,
	contribution_en = {Proposes a unified approach for generating 3D parts from point-to-mask geometric cues of varying granularity, enhancing cross-modal generation performance while providing flexibility through deep cross-modal fusion during training.},
	contribution_zh = {提出用从点到掩码的不同粒度几何提示统一实现三维部件的生成，通过在训练中的深度跨模态融合，在提供灵活性的同时提升了各模态的生成性能。}
}

MU-GeNeRF: Multi-view Uncertainty-guided Generalizable Neural Radiance Fields for Distractor-aware Scene

Wenjie Mu, Zhan Li, Chuanzhou Su, Xuanyi Shen, Ziniu Liu, Fan Lu, Yujian Mo, Junqiao Zhao*, Tiantian Feng, Chen Ye and Guang Chen

The IEEE/CVF Conference on Computer Vision and Pattern Recognition 2026, Denver, Colorado, 2026.

TL;DR: Proposes an uncertainty capture mechanism in generalizable NeRF for dynamic distractor scenes, enabling robust feedforward 3D reconstruction in such scenarios.

Link
BibTeX
@inproceedings{zotero-item-56854,
	author = {Mu, Wenjie and Li, Zhan and Su, Chuanzhou and Shen, Xuanyi and Liu, Ziniu and Lu, Fan and Mo, Yujian and Zhao*, Junqiao and Feng, Tiantian and Ye, Chen and Chen, Guang},
	title = {{{MU-GeNeRF}}: {{Multi-view Uncertainty-guided Generalizable Neural Radiance Fields}} for {{Distractor-aware Scene}}},
	booktitle = {The {{IEEE}}/{{CVF Conference}} on {{Computer Vision}} and {{Pattern Recognition}} 2026},
	year = 2026,
	address = {Denver, Colorado},
	url = {https://cvpr.thecvf.com/virtual/2026/poster/40235},
	contribution_en = {Proposes an uncertainty capture mechanism in generalizable NeRF for dynamic distractor scenes, enabling robust feedforward 3D reconstruction in such scenarios.},
	contribution_zh = {提出面向动态干扰场景的可泛化NeRF中的不确定性捕获机制，实现该类场景下的鲁棒前馈三维重建。}
}

Journal Articles

Ranking-aware Continual Learning for LiDAR Place Recognition

Xufei Wang, Gengxuan Tian, Junqiao Zhao*, Siyue Tao, Qiwen Gu, Qiankun You and Tiantian Feng

ISPRS Journal of Photogrammetry and Remote Sensing, vol. 236, pp. 197–211, 2026.

TL;DR: Introduces ranking-aware knowledge distillation and knowledge fusion to preserve LiDAR place-recognition performance as the model learns new environments.

DOI
BibTeX
@article{Xufei2025,
	author = {Wang, Xufei and Tian, Gengxuan and Zhao*, Junqiao and Tao, Siyue and Gu, Qiwen and You, Qiankun and Feng, Tiantian},
	title = {Ranking-aware Continual Learning for LiDAR Place Recognition},
	journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
	year = {2026},
	volume = {236},
	pages = {197--211},
	doi = {10.1016/j.isprsjprs.2026.03.037},
	contribution_en = {Introduces ranking-aware knowledge distillation and knowledge fusion to preserve LiDAR place-recognition performance as the model learns new environments.},
	contribution_zh = {提出排序感知的知识蒸馏损失和知识融合模块，帮助激光雷达地点识别模型在持续学习新环境时保留既有检索能力。},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

Learning From Mistakes: Loss-Aware Memory Enhanced Continual Learning for LiDAR Place Recognition

Xufei Wang, Junqiao Zhao*, Siyue Tao, Qiwen Gu, Wonbong Kim and Tiantian Feng

IEEE Robotics and Automation Letters, vol. 11, no. 8, pp. 9391–9398, 2026.

TL;DR: Introduces loss-aware memory replay for LiDAR place recognition, selectively revisiting samples prone to forgetting to preserve continual-learning performance.

DOI
BibTeX
@article{wangLearningMistakesLossAware,
	author = {Wang, Xufei and Zhao*, Junqiao and Tao, Siyue and Gu, Qiwen and Kim, Wonbong and Feng, Tiantian},
	title = {Learning {{From Mistakes}}: {{Loss-Aware Memory Enhanced Continual Learning}} for {{LiDAR Place Recognition}}},
	journal = {IEEE Robotics and Automation Letters},
	year = {2026},
	volume = {11},
	number = {8},
	pages = {9391--9398},
	doi = {10.1109/LRA.2026.3703260},
	contribution_en = {Introduces loss-aware memory replay for LiDAR place recognition, selectively revisiting samples prone to forgetting to preserve continual-learning performance.},
	contribution_zh = {提出面向激光雷达地点识别的损失感知记忆回放方法，通过有选择地重放易遗忘样本来保持持续学习性能。},
	shortjournal = {IEEE Robot. Autom. Lett.}
}

2025

Preprints and Manuscripts

ASTRO: Adaptive Stitching via Dynamics-Guided Trajectory Rollouts

Hang Yu, Di Zhang, Qiwei Du, Yanping Zhao, Hai Zhang, Guang Chen, Eduardo E. Veas and Junqiao Zhao*

arXiv, 2025.

Link
BibTeX
@article{Yu2026,
	author = {Hang Yu and Di Zhang and Qiwei Du and Yanping Zhao and Hai Zhang and Guang Chen and Eduardo E. Veas and Junqiao Zhao*},
	title = {ASTRO: Adaptive Stitching via Dynamics-Guided Trajectory Rollouts},
	journal = {arXiv},
	year = {2025},
	url = {https://arxiv.org/abs/2511.23442}
}

Deep Bundle-Adjusting Local-to-Global Registration for Unposed Generalizable Neural Radiance Fields

Wenjie Mu, Xuanyi Shen, Chuanzhou Su, Zhan Li, Junqiao Zhao*, Tiantian Feng, Chen Ye and Changjun Jiang

arXiv, 2025.

BibTeX
@article{Wenjie2025,
	author = {Mu, Wenjie and Shen, Xuanyi and Su, Chuanzhou and Li, Zhan and Zhao*, Junqiao and Feng, Tiantian and Ye, Chen and Jiang, Changjun},
	title = {Deep Bundle-Adjusting Local-to-Global Registration for Unposed Generalizable Neural Radiance Fields},
	journal = {arXiv},
	year = {2025},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

Multi-LVI-SAM: A Robust LiDAR-Visual-Inertial Odometry for Multiple Fisheye Cameras

Xinyu Zhang, Kai Huang, Junqiao Zhao*, Zihan Yuan and Tiantian Feng

arXiv, 2025.

Link
BibTeX
@article{Zhang2025,
	author = {Zhang, Xinyu and Huang, Kai and Zhao*, Junqiao and Yuan, Zihan and Feng, Tiantian},
	title = {Multi-LVI-SAM: A Robust LiDAR-Visual-Inertial Odometry for Multiple Fisheye Cameras},
	journal = {arXiv},
	year = {2025},
	url = {https://arxiv.org/abs/2509.05740},
	date-added = {2024-05-11 09:05:16 +0800},
	date-modified = {2024-05-11 09:41:54 +0800}
}

Conference Papers

Scrutinize What We Ignore: Reining Task Representation Shift In Context-Based Offline Meta Reinforcement Learning

Hai Zhang, Boyuan Zheng, Anqi Guo, Tianying Ji, Pheng-Ann Heng, Junqiao Zhao* and Lanqing Li

International Conference on Learning Representations (ICLR 2025), 2025.

TL;DR: Identifies task-representation shift caused by context-encoder updates and constrains those updates to stabilize context-based offline meta-RL.

BibTeX
@article{Reining2024,
	author = {Zhang, Hai and Zheng, Boyuan and Guo, Anqi and Ji, Tianying and Heng, Pheng-Ann and Zhao*, Junqiao and Li, Lanqing},
	title = {Scrutinize What We Ignore: Reining Task Representation Shift In Context-Based Offline Meta Reinforcement Learning},
	year = {2025},
	contribution_en = {Identifies task-representation shift caused by context-encoder updates and constrains those updates to stabilize context-based offline meta-RL.},
	contribution_zh = {分析上下文编码器更新引发的任务表征偏移，并通过约束更新稳定任务表示，改善上下文离线元强化学习。},
	date-added = {2024-08-12 19:44:04 +0800},
	date-modified = {2024-08-12 19:47:44 +0800},
	booktitle = {International Conference on Learning Representations (ICLR 2025)}
}

CERTAIN: Context Uncertainty-aware One-Shot Adaptation for Context-based Offline Meta Reinforcement Learning

Hongtu Zhou, Ruiling Yang, Yakun Zhu, Haoqi Zhao, Hai Zhang, Di Zhang, Junqiao Zhao*, Chen Ye and Changjun Jiang

International Conference on Machine Learning (ICML 2025), 2025.

TL;DR: Models uncertainty in context representations and uses it to selectively gather task context for improved one-shot adaptation in offline meta-RL.

Link
BibTeX
@inproceedings{Certain2025,
	author = {Zhou, Hongtu and Yang, Ruiling and Zhu, Yakun and Zhao, Haoqi and Zhang, Hai and Zhang, Di and Zhao*, Junqiao and Ye, Chen and Jiang, Changjun},
	title = {CERTAIN: Context Uncertainty-aware One-Shot Adaptation for Context-based Offline Meta Reinforcement Learning},
	booktitle = {International Conference on Machine Learning (ICML 2025)},
	year = {2025},
	url = {https://proceedings.mlr.press/v267/zhou25ai.html},
	contribution_en = {Models uncertainty in context representations and uses it to selectively gather task context for improved one-shot adaptation in offline meta-RL.},
	contribution_zh = {对上下文编码的不确定性进行建模，并据此选择性收集任务上下文，提升离线元强化学习的一次性适应能力。},
	date-added = {2025-06-12 19:44:04 +0800},
	date-modified = {2025-06-12 19:47:44 +0800}
}

Convex Hull-based Algebraic Constraint for Visual Quadric SLAM

Xiaolong Yu, Junqiao Zhao*, Shuangfu Song, Zhongyang Zhu, Zihan Yuan, Chen Ye, Tiantian Feng and Qiankun Yu

2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2025.

TL;DR: Uses convex-hull-based algebraic constraints in visual quadric SLAM to strengthen object-geometry cues for pose and scale estimation.

DOI
BibTeX
@inproceedings{Xiaolong2025,
	author = {Yu, Xiaolong and Zhao*, Junqiao and Song, Shuangfu and Zhu, Zhongyang and Yuan, Zihan and Ye, Chen and Feng, Tiantian and Yu, Qiankun},
	title = {Convex Hull-based Algebraic Constraint for Visual Quadric SLAM},
	booktitle = {2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
	year = {2025},
	doi = {10.1109/IROS60139.2025.11247367},
	contribution_en = {Uses convex-hull-based algebraic constraints in visual quadric SLAM to strengthen object-geometry cues for pose and scale estimation.},
	contribution_zh = {利用凸包构造视觉二次曲面 SLAM 的代数约束，增强物体几何约束对位姿与尺度估计的支持。},
	date-added = {2025-04-11 09:05:16 +0800},
	date-modified = {2025-04-11 09:41:54 +0800}
}

MutualVPR: A Mutual Learning Framework for Resolving Supervision Inconsistencies via Adaptive Clustering

Qiwen Gu, Xufei Wang, Junqiao Zhao*, Siyue Tao, Tiantian Feng, Ziqiao Wang and Guang Chen

Advances in Neural Information Processing Systems (NeurIPS 2025), 2025.

TL;DR: Introduces a mutual-learning place-recognition framework that jointly refines image descriptors and adaptive clusters to reduce supervision inconsistency from viewpoint changes and occlusion.

Link
BibTeX
@inproceedings{Qiwen2025,
	author = {Gu, Qiwen and Wang, Xufei and Zhao*, Junqiao and Tao, Siyue and Feng, Tiantian and Wang, Ziqiao and Chen, Guang},
	title = {MutualVPR: A Mutual Learning Framework for Resolving Supervision Inconsistencies via Adaptive Clustering},
	booktitle = {Advances in Neural Information Processing Systems (NeurIPS 2025)},
	year = {2025},
	url = {https://proceedings.neurips.cc/paper_files/paper/2025/hash/04388b6780bcf1a39ad0a91422f09a85-Abstract-Conference.html},
	contribution_en = {Introduces a mutual-learning place-recognition framework that jointly refines image descriptors and adaptive clusters to reduce supervision inconsistency from viewpoint changes and occlusion.},
	contribution_zh = {提出互学习地点识别框架，使图像特征学习与自适应聚类共同迭代，以缓解视角变化和遮挡造成的监督不一致。},
	date-added = {2025-05-11 09:05:16 +0800},
	date-modified = {2025-05-11 09:41:54 +0800}
}

CHPO: Constrained Hybrid-action Policy Optimization for Reinforcement Learning

Ao Zhou, Jiayi Guan, Li Shen, Fan Lu, Sanqing Qu, Junqiao Zhao, Ziqiao Wang, Ya Wu and Guang Chen

Advances in Neural Information Processing Systems (NeurIPS 2025), San Diego, USA, 2025.

TL;DR: Introduces constrained policy optimization for hybrid action spaces, with theoretical analysis of safe policy improvement.

Link
BibTeX
@inproceedings{zhouCHPOConstrainedHybridaction2025,
	author = {Zhou, Ao and Guan, Jiayi and Shen, Li and Lu, Fan and Qu, Sanqing and Zhao, Junqiao and Wang, Ziqiao and Wu, Ya and Chen, Guang},
	title = {{{CHPO}}: {{Constrained Hybrid-action Policy Optimization}} for {{Reinforcement Learning}}},
	booktitle = {Advances in Neural Information Processing Systems (NeurIPS 2025)},
	year = 2025,
	address = {San Diego, USA},
	url = {https://proceedings.neurips.cc/paper_files/paper/2025/hash/5eca2e4fe7858cbbfef4e08573cfcb25-Abstract-Conference.html},
	contribution_en = {Introduces constrained policy optimization for hybrid action spaces, with theoretical analysis of safe policy improvement.},
	contribution_zh = {提出适用于混合动作空间的约束策略优化方法，并建立安全策略改进的理论分析。}
}

Journal Articles

Batch Informed Vines (BIV*): Heuristically Guided Exploration of Narrow Passages by Batch Vine Expansion

Ziheng Jiang, Chang Huang, Hongtu Zhou, Linlin Zhou*, Chen Ye and Junqiao Zhao*

IEEE Robotics and Automation Letters, vol. 10, no. 2, pp. 1960–1967, 2025.

TL;DR: Introduces batch-informed Vine* search, using heuristic-guided batch expansion to improve exploration through narrow passages.

DOI
BibTeX
@article{Ziheng2024,
	author = {Jiang, Ziheng and Huang, Chang and Zhou, Hongtu and Zhou*, Linlin and Ye, Chen and Zhao*, Junqiao},
	title = {Batch Informed Vines (BIV*): Heuristically Guided Exploration of Narrow Passages by Batch Vine Expansion},
	journal = {IEEE Robotics and Automation Letters},
	year = {2025},
	volume = {10},
	number = {2},
	pages = {1960--1967},
	doi = {10.1109/LRA.2024.3518113},
	contribution_en = {Introduces batch-informed Vine* search, using heuristic-guided batch expansion to improve exploration through narrow passages.},
	contribution_zh = {提出批量知情的 Vine* 搜索方法，通过启发式引导的批量扩展提高狭窄通道中的路径探索效率。},
	date-modified = {2024-12-8 09:38:49 +0800}
}

UN3-Mapping: Uncertainty-Aware Neural Non-Projective Signed Distance Fields for 3D Mapping

Shuangfu Song, Junqiao Zhao*, Eduardo Veas, Jiaye Lin, Qiuyi Cao, Chen Ye and Tiantian Feng

IEEE Robotics and Automation Letters, pp. 1–8, 2025.

TL;DR: Develops an uncertainty-aware neural non-projective signed distance field mapping method that improves stability and accuracy in complex 3D reconstruction.

DOI
BibTeX
@article{songUN3MappingUncertaintyAwareNeural2025,
	author = {Song, Shuangfu and Zhao*, Junqiao and Veas, Eduardo and Lin, Jiaye and Cao, Qiuyi and Ye, Chen and Feng, Tiantian},
	title = {{{UN3-Mapping}}: {{Uncertainty-Aware Neural Non-Projective Signed Distance Fields}} for {{3D Mapping}}},
	journal = {IEEE Robotics and Automation Letters},
	year = {2025},
	pages = {1--8},
	doi = {10.1109/LRA.2025.3588410},
	contribution_en = {Develops an uncertainty-aware neural non-projective signed distance field mapping method that improves stability and accuracy in complex 3D reconstruction.},
	contribution_zh = {构建不确定性感知的神经非投影符号距离场建图方法，在复杂场景中提升三维重建稳定性与精度。},
	shorttitle = {{{UN3-Mapping}}},
	issn = {2377-3766},
	urldate = {2025-07-16}
}

2024

Conference Papers

Safe Reinforcement Learning with Dead-Ends Avoidance and Recovery

Xiao Zhang, Hai Zhang, Hongtu Zhou, Chang Huang, Di Zhang, Chen Ye* and Junqiao Zhao*

2024 IEEE International Conference on Robotics and Automation (ICRA), 2024.

TL;DR: Separates task and recovery policies and uses safety-value assessment with online correction to reduce dead-end risks in reinforcement learning.

BibTeX
@inproceedings{ZhangXiao-2023,
	author = {Zhang, Xiao and Zhang, Hai and Zhou, Hongtu and Huang, Chang and Zhang, Di and Ye*, Chen and Zhao*, Junqiao},
	title = {Safe Reinforcement Learning with Dead-Ends Avoidance and Recovery},
	booktitle = {2024 IEEE International Conference on Robotics and Automation (ICRA)},
	year = {2024},
	contribution_en = {Separates task and recovery policies and uses safety-value assessment with online correction to reduce dead-end risks in reinforcement learning.},
	contribution_zh = {将安全策略与恢复策略分开学习，并利用安全价值评估和在线纠正降低强化学习进入死区的风险。},
	date-added = {2023-09-06 14:20:28 +0800},
	date-modified = {2024-05-11 09:40:01 +0800}
}

LOG-LIO: A LiDAR-Inertial Odometry with Efficient Local Geometric Information Estimation

Kai Huang, Junqiao Zhao*, Zhongyang Zhu, Chen Ye and Tiantian Feng

2024 IEEE International Conference on Robotics and Automation (ICRA), 2024.

TL;DR: Introduces efficient local geometric-information estimation for tightly coupled LiDAR-inertial odometry.

BibTeX
@inproceedings{Huangkai2023,
	author = {Huang, Kai and Zhao*, Junqiao and Zhu, Zhongyang and Ye, Chen and Feng, Tiantian},
	title = {LOG-LIO: A LiDAR-Inertial Odometry with Efficient Local Geometric Information Estimation},
	booktitle = {2024 IEEE International Conference on Robotics and Automation (ICRA)},
	year = {2024},
	contribution_en = {Introduces efficient local geometric-information estimation for tightly coupled LiDAR-inertial odometry.},
	contribution_zh = {提出高效局部几何信息估计方法，结合激光雷达与惯性测量实现紧耦合里程计。},
	date-added = {2023-09-06 14:21:46 +0800},
	date-modified = {2024-05-11 09:40:08 +0800}
}

N$^3$-Mapping: Normal Guided Neural Non-Projective Signed Distance Fields for Large-scale 3D Mapping

Shuangfu Song, Junqiao Zhao*, Kai Huang, Jiaye Lin, Chen Ye and Tiantian Feng

2024 IEEE International Conference on Robotics and Automation (ICRA), Yokohama, Japan, 2024.

TL;DR: Uses surface-normal constraints in neural non-projective signed distance fields, with voxel-based sliding windows and hierarchical sampling for efficient large-scale mapping.

BibTeX
@conference{Song2024,
	author = {Song, Shuangfu and Zhao*, Junqiao and Huang, Kai and Lin, Jiaye and Ye, Chen and Feng, Tiantian},
	title = {N$^{3}$-Mapping: Normal Guided Neural Non-Projective Signed Distance Fields for Large-scale 3D Mapping},
	booktitle = {2024 IEEE International Conference on Robotics and Automation (ICRA)},
	year = {2024},
	address = {Yokohama, Japan},
	contribution_en = {Uses surface-normal constraints in neural non-projective signed distance fields, with voxel-based sliding windows and hierarchical sampling for efficient large-scale mapping.},
	contribution_zh = {利用表面法向约束神经非投影符号距离场，并结合体素滑动窗口与分层采样实现高效的大规模三维建图。},
	date-added = {2024-05-11 08:54:37 +0800},
	date-modified = {2024-05-11 09:39:51 +0800}
}

Focus On What Matters: Separated Models For Visual-Based RL Generalization

Di Zhang, Bowen Lv, Hai Zhang, Feifan Yang, Hongtu Zhou, Chang Huang, Hang Yu, Chen Ye, Junqiao Zhao* and Changjun Jiang

Advances in Neural Information Processing Systems (NeurIPS 2024), 2024.

TL;DR: Separates task-relevant and task-irrelevant visual factors in reinforcement learning and uses consistency constraints to improve policy generalization.

BibTeX
@article{SMG2024,
	author = {Zhang, Di and Lv, Bowen and Zhang, Hai and Yang, Feifan and Zhou, Hongtu and Huang, Chang and Yu, Hang and Ye, Chen and Zhao*, Junqiao and Jiang, Changjun},
	title = {Focus On What Matters: Separated Models For Visual-Based RL Generalization},
	year = {2024},
	contribution_en = {Separates task-relevant and task-irrelevant visual factors in reinforcement learning and uses consistency constraints to improve policy generalization.},
	contribution_zh = {将视觉强化学习中的任务相关与无关因素分开建模，并通过一致性约束提升策略的泛化能力。},
	date-added = {2024-08-12 19:47:54 +0800},
	date-modified = {2024-08-12 19:49:55 +0800},
	booktitle = {Advances in Neural Information Processing Systems (NeurIPS 2024)}
}

Towards an Information Theoretic Framework of Context-Based Offline Meta-Reinforcement Learning

Lanqing Li, Hai Zhang, Xinyu Zhang, Shatong Zhu, Junqiao Zhao* and Pheng-Ann Heng

Advances in Neural Information Processing Systems (NeurIPS 2024), Spotlight, 2024.

TL;DR: Develops an information-theoretic account of task information in context-based offline meta-RL and uses it to analyze context representation and policy learning.

BibTeX
@article{Unicorn2024,
	author = {Li, Lanqing and Zhang, Hai and Zhang, Xinyu and Zhu, Shatong and Zhao*, Junqiao and Heng, Pheng-Ann},
	title = {Towards an Information Theoretic Framework of Context-Based Offline Meta-Reinforcement Learning},
	year = {2024},
	contribution_en = {Develops an information-theoretic account of task information in context-based offline meta-RL and uses it to analyze context representation and policy learning.},
	contribution_zh = {从信息论角度刻画上下文离线元强化学习中的任务信息，并据此分析上下文表示与策略学习。},
	date-modified = {2024-05-11 09:42:06 +0800},
	booktitle = {Advances in Neural Information Processing Systems (NeurIPS 2024), Spotlight}
}

Journal Articles

Safe Reinforcement Learning with Dead-Ends Avoidance and Recovery

Xiao Zhang, Hai Zhang, Hongtu Zhou, Chang Huang, Di Zhang, Chen Ye* and Junqiao Zhao*

IEEE Robotics and Automation Letters, vol. 9, no. 1, pp. 491–498, 2024.

TL;DR: Separates task and safety-recovery policies and uses safety assessment with online correction to avoid unrecoverable unsafe states in reinforcement learning.

DOI
BibTeX
@article{ZhangXiao-2023,
	author = {Zhang, Xiao and Zhang, Hai and Zhou, Hongtu and Huang, Chang and Zhang, Di and Ye*, Chen and Zhao*, Junqiao},
	title = {Safe Reinforcement Learning with Dead-Ends Avoidance and Recovery},
	journal = {IEEE Robotics and Automation Letters},
	year = {2024},
	volume = {9},
	number = {1},
	pages = {491--498},
	doi = {10.1109/LRA.2023.3333248},
	contribution_en = {Separates task and safety-recovery policies and uses safety assessment with online correction to avoid unrecoverable unsafe states in reinforcement learning.},
	contribution_zh = {将任务策略与安全恢复策略分开，并通过安全评估和在线纠正避免强化学习进入不可恢复的危险状态。},
	date-added = {2023-09-06 14:20:28 +0800},
	date-modified = {2023-09-06 14:21:38 +0800}
}

LOG-LIO: A LiDAR-Inertial Odometry with Efficient Local Geometric Information Estimation

Kai Huang, Junqiao Zhao*, Zhongyang Zhu, Chen Ye and Tiantian Feng

IEEE Robotics and Automation Letters, vol. 9, no. 1, pp. 459-466, 2024.

TL;DR: Introduces efficient local geometric-information estimation for state estimation in LiDAR-inertial odometry.

DOI
BibTeX
@article{Huangkai2023,
	author = {Huang, Kai and Zhao*, Junqiao and Zhu, Zhongyang and Ye, Chen and Feng, Tiantian},
	title = {LOG-LIO: A LiDAR-Inertial Odometry with Efficient Local Geometric Information Estimation},
	journal = {IEEE Robotics and Automation Letters},
	year = {2024},
	volume = {9},
	number = {1},
	pages = {459-466},
	doi = {10.1109/LRA.2023.3332020},
	contribution_en = {Introduces efficient local geometric-information estimation for state estimation in LiDAR-inertial odometry.},
	contribution_zh = {提出局部几何信息高效估计方法，并将其用于激光雷达惯性里程计中的状态估计。},
	date-added = {2023-09-06 14:21:46 +0800},
	date-modified = {2023-09-06 14:23:42 +0800}
}

DL-SLOT: Tightly-Coupled Dynamic LiDAR SLAM and 3D Object Tracking Based on Collaborative Graph Optimization

Xuebo Tian, Zhongyang Zhu, Junqiao Zhao*, Gengxuan Tian and Chen Ye

IEEE Transactions on Intelligent Vehicles, vol. 9, no. 1, pp. 1017–1027, 2024.

TL;DR: Jointly estimates ego pose and 3D object tracks through collaborative graph optimization, coupling localization and tracking in dynamic scenes.

DOI
BibTeX
@article{Tian_DLSLOT_2023,
	author = {Tian, Xuebo and Zhu, Zhongyang and Zhao*, Junqiao and Tian, Gengxuan and Ye, Chen},
	title = {DL-SLOT: Tightly-Coupled Dynamic LiDAR SLAM and 3D Object Tracking Based on Collaborative Graph Optimization},
	journal = {IEEE Transactions on Intelligent Vehicles},
	year = {2024},
	volume = {9},
	number = {1},
	pages = {1017--1027},
	doi = {10.1109/TIV.2023.3317308},
	contribution_en = {Jointly estimates ego pose and 3D object tracks through collaborative graph optimization, coupling localization and tracking in dynamic scenes.},
	contribution_zh = {在协同图优化中联合估计自车位姿与三维目标轨迹，解除静态环境和已知自车位姿假设，实现动态场景下的紧耦合定位与跟踪。},
	date-added = {2023-09-26 20:17:56 +0800},
	date-modified = {2023-09-26 20:21:43 +0800}
}

Learning Sequence Descriptor Based on Spatio-Temporal Attention for Visual Place Recognition

Junqiao Zhao, Fenglin Zhang, Yingfeng Cai, Gengxuan Tian, Wenjie Mu, Chen Ye and Tiantian Feng

IEEE Robotics and Automation Letters, vol. 9, no. 3, pp. 2351–2358, 2024.

TL;DR: Aggregates image-sequence information with spatiotemporal attention to learn sequence-level descriptors for visual place recognition.

DOI
BibTeX
@article{ZhangFL2024,
	author = {Zhao, Junqiao and Zhang, Fenglin and Cai, Yingfeng and Tian, Gengxuan and Mu, Wenjie and Ye, Chen and Feng, Tiantian},
	title = {Learning Sequence Descriptor Based on Spatio-Temporal Attention for Visual Place Recognition},
	journal = {IEEE Robotics and Automation Letters},
	year = {2024},
	volume = {9},
	number = {3},
	pages = {2351--2358},
	doi = {10.1109/LRA.2024.3354627},
	contribution_en = {Aggregates image-sequence information with spatiotemporal attention to learn sequence-level descriptors for visual place recognition.},
	contribution_zh = {通过时空注意力聚合图像序列信息，学习用于视觉地点识别的序列级描述子。},
	date-added = {2023-09-06 14:18:02 +0800},
	date-modified = {2023-09-06 14:20:20 +0800}
}

N$^3$-Mapping: Normal Guided Neural Non-Projective Signed Distance Fields for Large-scale 3D Mapping

Shuangfu Song, Junqiao Zhao*, Kai Huang, Jiaye Lin, Chen Ye and Tiantian Feng

IEEE Robotics and Automation Letters, vol. 9, no. 6, pp. 5935–5942, 2024.

TL;DR: Uses surface-normal constraints in neural non-projective signed distance fields, with voxel-based sliding windows and hierarchical sampling for efficient large-scale mapping.

DOI
BibTeX
@article{Song2024,
	author = {Song, Shuangfu and Zhao*, Junqiao and Huang, Kai and Lin, Jiaye and Ye, Chen and Feng, Tiantian},
	title = {N$^{3}$-Mapping: Normal Guided Neural Non-Projective Signed Distance Fields for Large-scale 3D Mapping},
	journal = {IEEE Robotics and Automation Letters},
	year = {2024},
	volume = {9},
	number = {6},
	pages = {5935--5942},
	doi = {10.1109/LRA.2024.3396638},
	contribution_en = {Uses surface-normal constraints in neural non-projective signed distance fields, with voxel-based sliding windows and hierarchical sampling for efficient large-scale mapping.},
	contribution_zh = {利用表面法向约束神经非投影符号距离场，并结合体素滑动窗口与分层采样实现高效的大规模三维建图。},
	date-added = {2024-05-11 08:54:09 +0800},
	date-modified = {2024-05-11 08:54:24 +0800}
}

LIMOT: A Tightly-Coupled System for LiDAR-Inertial Odometry and Multi-Object Tracking

Zhongyang Zhu, Junqiao Zhao*, Kai Huang, Xuebo Tian, Jiaye Lin and Chen Ye

IEEE Robotics and Automation Letters, vol. 9, no. 7, pp. 6600–6607, 2024.

TL;DR: Tightly couples LiDAR-inertial odometry and multi-object tracking, jointly estimating ego-vehicle and surrounding-object poses in a factor graph.

DOI
BibTeX
@article{ZhuZY2023,
	author = {Zhu, Zhongyang and Zhao*, Junqiao and Huang, Kai and Tian, Xuebo and Lin, Jiaye and Ye, Chen},
	title = {LIMOT: A Tightly-Coupled System for LiDAR-Inertial Odometry and Multi-Object Tracking},
	journal = {IEEE Robotics and Automation Letters},
	year = {2024},
	volume = {9},
	number = {7},
	pages = {6600--6607},
	doi = {10.1109/LRA.2024.3402628},
	contribution_en = {Tightly couples LiDAR-inertial odometry and multi-object tracking, jointly estimating ego-vehicle and surrounding-object poses in a factor graph.},
	contribution_zh = {将激光雷达惯性里程计与多目标跟踪紧密耦合，在因子图中联合估计自车和周围运动目标位姿。},
	date-added = {2024-05-15 13:23:05 +0800},
	date-modified = {2024-05-15 13:24:06 +0800}
}

LOG-LIO2: A LiDAR-Inertial Odometry with Efficient Uncertainty Analysis

Kai Huang, Junqiao Zhao*, Jiaye Lin, Zhongyang Zhu, Shuangfu Song, Chen Ye and Tiantian Feng

IEEE Robotics and Automation Letters, vol. 9, no. 10, pp. 8226–8233, 2024.

TL;DR: Models LiDAR point uncertainty from range, incidence angle, and surface properties, and reduces propagation cost with an incremental local approximation.

DOI
BibTeX
@article{Loglio22024,
	author = {Huang, Kai and Zhao*, Junqiao and Lin, Jiaye and Zhu, Zhongyang and Song, Shuangfu and Ye, Chen and Feng, Tiantian},
	title = {LOG-LIO2: A LiDAR-Inertial Odometry with Efficient Uncertainty Analysis},
	journal = {IEEE Robotics and Automation Letters},
	year = {2024},
	volume = {9},
	number = {10},
	pages = {8226--8233},
	doi = {10.1109/LRA.2024.3440850},
	contribution_en = {Models LiDAR point uncertainty from range, incidence angle, and surface properties, and reduces propagation cost with an incremental local approximation.},
	contribution_zh = {提出综合考虑测距、入射角和表面特性的点不确定性模型，并以局部增量近似降低不确定性传播的计算开销。},
	date-added = {2024-08-12 19:43:54 +0800},
	date-modified = {2024-08-13 13:39:10 +0800}
}

VNI-Net: Vector Neurons-based Rotation-Invariant Descriptor for LiDAR Place Recognition

Gengxuan Tian, Junqiao Zhao*, Yingfeng Cai, Fenglin Zhang, Xufei Wang, Wenjie Mu, Chen Ye and Sisi Zlatanova

ISPRS Journal of Photogrammetry and Remote Sensing, vol. 218, pp. 506–517, 2024.

TL;DR: Uses vector neurons to construct rotation-invariant LiDAR place descriptors for robust recognition across orientations.

DOI
BibTeX
@article{TianGX2023,
	author = {Tian, Gengxuan and Zhao*, Junqiao and Cai, Yingfeng and Zhang, Fenglin and Wang, Xufei and Mu, Wenjie and Ye, Chen and Zlatanova, Sisi},
	title = {VNI-Net: Vector Neurons-based Rotation-Invariant Descriptor for LiDAR Place Recognition},
	journal = {ISPRS Journal of Photogrammetry and Remote Sensing},
	year = {2024},
	volume = {218},
	pages = {506--517},
	doi = {10.1016/j.isprsjprs.2024.09.011},
	contribution_en = {Uses vector neurons to construct rotation-invariant LiDAR place descriptors for robust recognition across orientations.},
	contribution_zh = {利用向量神经元构造旋转不变的激光雷达地点描述子，提升不同朝向下地点识别的稳健性。},
	date-modified = {2024-05-11 09:38:49 +0800}
}

2023

Conference Papers

Multi-agent Decision-making at Unsignalized Intersections with Reinforcement Learning from Demonstrations

Chang Huang, Junqiao Zhao*, Hongtu Zhou, Hai Zhang, Xiao Zhang and Chen Ye

2023 IEEE Intelligent Vehicles Symposium (IV), Anchorage, Alaska, IEEE IV2023, 2023.

TL;DR: Combines learning from demonstrations with reinforcement learning for cooperative multi-agent decision-making at unsignalized intersections.

BibTeX
@conference{IV-2023chang,
	author = {Huang, Chang and Zhao*, Junqiao and Zhou, Hongtu and Zhang, Hai and Zhang, Xiao and Ye, Chen},
	title = {Multi-agent Decision-making at Unsignalized Intersections with Reinforcement Learning from Demonstrations},
	booktitle = {2023 IEEE Intelligent Vehicles Symposium (IV)},
	year = {2023},
	address = {Anchorage, Alaska},
	publisher = {IEEE IV2023},
	contribution_en = {Combines learning from demonstrations with reinforcement learning for cooperative multi-agent decision-making at unsignalized intersections.},
	contribution_zh = {结合示范学习与强化学习，为无信号交叉口的多智能体车辆提供协同决策方法。}
}

How to Fine-tune the Model: Unified Model Shift and Model Bias Policy Optimization

Hai Zhang, Hang Yu, Junqiao Zhao*, Di Zhang, Chang Huang, Hongtu Zhou, Xiao Zhang and Chen Ye

Advances in Neural Information Processing Systems (NeurIPS 2023), 2023.

TL;DR: Unifies model shift and model bias in offline reinforcement learning and adaptively fine-tunes models to improve policy performance.

BibTeX
@conference{Zhang-finetune2023,
	author = {Zhang, Hai and Yu, Hang and Zhao*, Junqiao and Zhang, Di and Huang, Chang and Zhou, Hongtu and Zhang, Xiao and Ye, Chen},
	title = {How to Fine-tune the Model: Unified Model Shift and Model Bias Policy Optimization},
	booktitle = {Advances in Neural Information Processing Systems (NeurIPS 2023)},
	year = {2023},
	contribution_en = {Unifies model shift and model bias in offline reinforcement learning and adaptively fine-tunes models to improve policy performance.},
	contribution_zh = {统一处理模型偏移与模型偏差，并通过自适应微调提升离线强化学习策略的改进效果。},
	date-added = {2023-09-06 14:10:32 +0800},
	date-modified = {2023-09-26 20:17:48 +0800}
}

2022

Conference Papers

Scale Estimation with Dual Quadrics for Monocular Object SLAM

Shuangfu Song, Junqiao Zhao*, Tiantian Feng, Chen Ye and Lu Xiong

2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Kyoto, Japan, IEEE IROS2022, 2022.

TL;DR: Estimates absolute scale in monocular object SLAM using dual-quadric object models and physical-dimension priors.

BibTeX
@conference{IROS-2022ssf,
	author = {Song, Shuangfu and Zhao*, Junqiao and Feng, Tiantian and Ye, Chen and Xiong, Lu},
	title = {Scale Estimation with Dual Quadrics for Monocular Object SLAM},
	booktitle = {2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
	year = {2022},
	address = {Kyoto, Japan},
	publisher = {IEEE IROS2022},
	contribution_en = {Estimates absolute scale in monocular object SLAM using dual-quadric object models and physical-dimension priors.},
	contribution_zh = {利用物体的双二次曲面模型及尺寸先验，从单目物体 SLAM 中估计绝对尺度。}
}

UPC-Faster-RCNN: A Dynamic Self-Labeling Algorithm for Open-Set Object Detection Based on Unknown Proposal Clustering

Yujun Liao, Yan Wu, Yujian Mo, Feilin Liu, Yufei He and Junqiao Zhao

2022 IEEE International Conference on Multisensor Fusion and Integration (MFI), pp. 1–6, Bedford, UK, IEEE, 2022.

TL;DR: No public contribution summary is available.

DOI
BibTeX
@conference{MFI-2022lyj,
	author = {Liao, Yujun and Wu, Yan and Mo, Yujian and Liu, Feilin and He, Yufei and Zhao, Junqiao},
	title = {UPC-Faster-RCNN: A Dynamic Self-Labeling Algorithm for Open-Set Object Detection Based on Unknown Proposal Clustering},
	booktitle = {2022 IEEE International Conference on Multisensor Fusion and Integration (MFI)},
	year = {2022},
	pages = {1--6},
	address = {Bedford, UK},
	publisher = {IEEE},
	doi = {10.1109/MFI55806.2022.9913864}
}

Patch-NetVLAD+: Learned patch descriptor and weighted matching strategy for place recognition

Yingfeng Cai, Junqiao Zhao*, Jiafeng Cui, Fenglin Zhang, Tiantian Feng and Chen Ye

2022 IEEE International Conference on Multisensor Fusion and Integration (MFI), Cranfield, UK, IEEE MFI2022, 2022.

TL;DR: Improves visual place recognition by learning local patch descriptors and weighting their matches.

BibTeX
@conference{MFI-2022cyf,
	author = {Cai, Yingfeng and Zhao*, Junqiao and Cui, Jiafeng and Zhang, Fenglin and Feng, Tiantian and Ye, Chen},
	title = {Patch-NetVLAD+: Learned patch descriptor and weighted matching strategy for place recognition},
	booktitle = {2022 IEEE International Conference on Multisensor Fusion and Integration (MFI)},
	year = {2022},
	address = {Cranfield, UK},
	publisher = {IEEE MFI2022},
	contribution_en = {Improves visual place recognition by learning local patch descriptors and weighting their matches.},
	contribution_zh = {通过学习局部图像块描述子并对匹配结果加权，提升视觉地点识别的检索效果。}
}

Journal Articles

LIO-Vehicle: A Tightly-Coupled Vehicle Dynamics Extension of Lidar Inertial Odometry

Hongru Xiao, Yanqun Han, Junqiao Zhao*, Jiafeng Cui, Lu Xiong and Zhuoping Yu

IEEE Robotics and Automation Letters (RAL), vol. 7, pp. 446-453, IEEE, 2022.

TL;DR: Tightly integrates a vehicle-dynamics model into LiDAR-inertial odometry to improve motion estimation for autonomous vehicles.

DOI
BibTeX
@article{RAL-2021Xhr,
	author = {Xiao, Hongru and Han, Yanqun and Zhao*, Junqiao and Cui, Jiafeng and Xiong, Lu and Yu, Zhuoping},
	title = {LIO-Vehicle: A Tightly-Coupled Vehicle Dynamics Extension of Lidar Inertial Odometry},
	journal = {IEEE Robotics and Automation Letters (RAL)},
	year = {2022},
	volume = {7},
	pages = {446-453},
	doi = {10.1109/LRA.2021.3126336},
	contribution_en = {Tightly integrates a vehicle-dynamics model into LiDAR-inertial odometry to improve motion estimation for autonomous vehicles.},
	contribution_zh = {将车辆动力学模型紧耦合到激光雷达惯性里程计，提升自主车辆的运动估计精度。},
	issn = {2377-3766},
	issue = {1},
	publisher = {IEEE},
	bdsk-url-1 = {https://doi.org/10.1109/LRA.2021.3126336}
}

融合车辆动力学的双目视觉惯性SLAM研究

余卓平, 鞠然, 韩燕群 and 赵君峤*

华中科技大学学报（自然科学版）, 2022.

TL;DR: Integrates vehicle-dynamics constraints into stereo visual-inertial SLAM to improve state estimation under degenerate ground-vehicle motion.

DOI
BibTeX
@article{HUST-2022Juran,
	author = {余卓平 and 鞠然 and 韩燕群 and 赵君峤*},
	title = {融合车辆动力学的双目视觉惯性SLAM研究},
	journal = {华中科技大学学报（自然科学版）},
	year = {2022},
	doi = {10.13245/j.hust.220509},
	contribution_en = {Integrates vehicle-dynamics constraints into stereo visual-inertial SLAM to improve state estimation under degenerate ground-vehicle motion.},
	contribution_zh = {将车辆动力学约束融入双目视觉惯性 SLAM，以改善地面车辆运动退化条件下的状态估计。},
	bdsk-url-1 = {https://doi.org/10.13245/j.hust.220509}
}

G-VIDO: A Vehicle Dynamics and Intermittent GNSS Aided Visual-Inertial State Estimator For Autonomous Driving

Lu Xiong, Rong Kang, Junqiao Zhao*, Peizhi Zhang, Mingyu Xu, Ran Ju, Chen Ye and Tiantian Feng

IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 8, pp. 11845-11861, 2022.

TL;DR: Combines vehicle-dynamics constraints and intermittent GNSS with visual-inertial estimation, using anomaly detection for robust localization across GNSS conditions.

DOI
BibTeX
@article{IEEEITS-2021Kf,
	author = {Xiong, Lu and Kang, Rong and Zhao*, Junqiao and Zhang, Peizhi and Xu, Mingyu and Ju, Ran and Ye, Chen and Feng, Tiantian},
	title = {G-VIDO: A Vehicle Dynamics and Intermittent GNSS Aided Visual-Inertial State Estimator For Autonomous Driving},
	journal = {IEEE Transactions on Intelligent Transportation Systems},
	year = {2022},
	volume = {23},
	number = {8},
	pages = {11845-11861},
	doi = {10.1109/TITS.2021.3107873},
	contribution_en = {Combines vehicle-dynamics constraints and intermittent GNSS with visual-inertial estimation, using anomaly detection for robust localization across GNSS conditions.},
	contribution_zh = {将车辆动力学约束与间歇式 GNSS 融入视觉惯性状态估计，并通过异常检测提升不同 GNSS 条件下的定位连续性与鲁棒性。},
	bdsk-url-1 = {https://doi.org/10.1109/TITS.2021.3107873}
}

Weighted octree-based 3D indoor pathfinding for multiple locomotion types

Junqiao Zhao, Qiaoyu Xu, Sisi Zlatanova, Liu Liu, Chen Ye and Tiantian Feng

International Journal of Applied Earth Observation and Geoinformation, vol. 112, pp. 102900, 2022.

TL;DR: Defines traversal costs for different locomotion types in a multiresolution octree to support indoor path planning for multiple robot types.

LinkDOI
BibTeX
@article{ZHAO2022102900,
	author = {Junqiao Zhao and Qiaoyu Xu and Sisi Zlatanova and Liu Liu and Chen Ye and Tiantian Feng},
	title = {Weighted octree-based 3D indoor pathfinding for multiple locomotion types},
	journal = {International Journal of Applied Earth Observation and Geoinformation},
	year = {2022},
	volume = {112},
	pages = {102900},
	url = {https://www.sciencedirect.com/science/article/pii/S1569843222001029},
	doi = {https://doi.org/10.1016/j.jag.2022.102900},
	contribution_en = {Defines traversal costs for different locomotion types in a multiresolution octree to support indoor path planning for multiple robot types.},
	contribution_zh = {在多分辨率八叉树中为不同移动方式定义通行代价，从而支持室内环境中的多类型机器人路径规划。},
	issn = {1569-8432},
	bdsk-url-1 = {https://www.sciencedirect.com/science/article/pii/S1569843222001029},
	bdsk-url-2 = {https://doi.org/10.1016/j.jag.2022.102900}
}