|
Yue Chen | 陈越 I am currently a graduate student at Peking University with Agibot Lab advised by Professor Hao Dong. I am also fortunate to have mentorship from Ruihai Wu. I've also had great experiences working at MSRA and Seed ByteDance. My research interest is broadly in Robotics, 3D Computer Vision and large language models (LLMs), with particular interests in generalizable object manipulation. |
|
News
|
ResearchI'm open to collaborations on related projects, feel free to contact me! Papers sorted by recency. |
|
Learning Part-Aware Dense 3D Feature Field For Generalizable Articulated Object Manipulation
, , , , , , , project page / paper / code
Under Review
TL;DR:
We propose PA3FF (Part-Aware 3D Feature Field), a 3D-native dense feature representation that encodes functional part awareness directly from point clouds. Combined with PADP (Part-Aware Diffusion Policy), it enhances sample efficiency and generalization in articulated object manipulation—outperforming 2D/3D baselines (CLIP, DINOv2, Grounded-SAM) on 16 simulated (PartInstruct) and 8 real-world tasks, and enabling downstream tasks like 3D shape correspondence and part segmentation.
|
|
DexGarmentLab: Dexterous Garment Manipulation Environment with Generalizable Policy
*, *, *, , , , , , , project page / paper / code / data
TL;DR:
We introduce DexGarmentLab, a realistic sim environment for bimanual dexterous garment manipulation. Based on this environment, we propose a new benchmark, an efficient data collection pipeline, and a novel policy framework that uses category-level visual correspondences for few-shot garment manipulation.
|
|
ExeCoder: Empowering Large Language Models with Executability Representation for Code Translation
, , , , , , , , paper / project page / code
TL;DR:
We introduce ExeCoder, an LLM specifically designed for code translation, aimed at utilizing executability representations such as functional semantics, syntax structures, and variable dependencies to enhance the capabilities of LLMs in code translation.
|
|
Duet: Joint Exploration of User–Item Profiles
, , , , , , , , , , , , , , , , , , project page / paper / code
Under Review
TL;DR:
We propose DUET, a closed-loop framework for joint exploration of user-item textual profiles in recommendation systems. It distills raw data into concise cues, expands them into rich profiles via self-prompt construction, and optimizes profiles jointly with reinforcement learning using downstream recommendation feedback, while enabling interpretable LLM-compatible representations.
|
|
WarriorMath: Enhancing the Mathematical Ability of Large Language Models with a Defect-aware Framework
, , , , , , , , , , , paper / project page / code
Under Review
TL;DR:
We propose WarriorMath, a defect-aware framework for mathematical problem solving that integrates both targeted data synthesis and progressive training.
|
|
TrustRAG: Enhancing Robustness and Trustworthiness in RAG
*, *, *, , , , , project page / paper / code
Under Review
TL;DR:
We introduce TrustRAG, a robust Retrieval-Augmented Generation (RAG) framework. It defends against corpus poisoning attacks by a two-stage mechanism: identifying potential attack patterns with K-means clustering and detecting malicious docs via self-assessment.
|
|
Point-Level Visual Affordance Guided Retrieval and Adaptation for Cluttered Garments Manipulation
*, *, *, , , project page / paper / code
CVPR 2025
Computer Vision and Pattern Recognition
TL;DR:
We study the novel task of cluttered garments manipulation using dense visual affordance, with generalization towards diverse states, and propose a novel adaptation module to reorganize cluttered garments into configurations conducive to manipulation.
|
|
ET-SEED: Efficient Trajectory-Level SE(3) Equivariant Diffusion Policy
*, *, *, , , , project page / paper / code / video
ICLR 2025
International Conference on Learning Representations
TL;DR:
We theoretically extend equivariant Markov kernels and simplify the condition of equivariant diffusion process, thereby significantly improving training efficiency for trajectory-level SE(3) equivariant diffusion policy in an end-to-end manner.
|
|
EqvAfford: SE(3) Equivariance for Point-Level Affordance Learning
*, *, *, paper
CVPR 2024 Workshop EquiVision
Computer Vision and Pattern Recognition
TL;DR:
We propose EqvAfford framework, with novel designs to guarantee the SE(3) equivariance in point-level affordance learning for downstream robotic manipulation.
|
Honors and Awards
|
|
Thank you for visiting! Feel free to contact me if you have any questions.
|