- Incentivizing Multimodal Reasoning in Large Models for Direct Robot Manipulation, submitted to NeurIPS 2025
- Geometric Constraints as General Interface for Robotic Manipulations, submitted to ICML 2025
- Embodiment-Agnostic Action Planning via Object-Part Scene Flow, accepted by ICRA 2025
- Moto: Latent Motion Token as the Bridging Language for Robot Manipulation, submitted to ICCV 2025
- Overcoming Support Dilution for Robust Few-shot Semantic Segmentation, submitted to TNNLS 2025
- Prototypical Variational Autoencoder for Few-shot 3D Point Cloud Object Detection, accepted by NeurIPS 2025
- SKU-Patch: Towards Efficient Instance Segmentation for Unseen Objects in Auto-Store, submitted to Arxiv 2023
- SKU-Patch: Towards Efficient Instance Segmentation for Unseen Objects in Auto-Store, accepted by AAAI 2021
- SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloud, accepted by CVPR 2021
Research Experience
- Project: Virtual Training Data Generation Pipeline with Unreal Engine 4, 2021
Education
- Ph.D. in Computer Science and Engineering (CSE), The Chinese University of Hong Kong (CUHK), supervised by Prof. Chi-Wing Fu, 2021-present
- B. Eng. from The Chinese University of Hong Kong (CUHK), 2021
Background
- Research Interests: Embodiment AI, achieving robot manipulation given only language description and generalizable to any tasks, any robots, any objects, and any scenarios.
- Professional Fields: 3D computer vision, autonomous driving, few-shot learning.
Miscellany
- Teaching Experience:
- 2021-2022 Fall: Introduction to Algorithm (CSCI 3160)
- 2021-2022 Spring: Principle of Software Engineering (CSCI 3180)