Xinyue Zhu
Email: xz3013 [at] columbia [dot] edu
I'm an undergraduate student at Columbia University. At Columbia, I am fortunate to work with Prof. Yunzhu Li and Prof. Matei Ciocarlie.
My research interest is in robot learning. I'm broadly interested in developing structured and scalable learning methods that enable robots to understand and interact reliably with the physical world.
Best Demo Awardat RSS 2025 Workshop on Robot Hardware-Aware Intelligence [Link]
Conference on Neural Information Processing Systems (NeurIPS), 2025
Developed a portable visuo-tactile gripper and a 2.6M-sample in-the-wild dataset for fine-grained manipulation.
Under Review, 2025
Developed a masked-modeling transformer for fast, zero-shot intent detection in wearable orthosis control.
Autonomous Robots Journal, 2025
Trained end-to-end RL policies with high-fidelity tactile feedback to retrieve objects buried in granular media.
Workshop on Interpreting Cognition in Deep Learning Models, NeurIPS 2025 [Link]
Under Review, 2025
Uncovered interpretable dynamics underlying human generalization via an RNN with sparse latent representations.
IEEE Robotics and Automation Letters(RA-L), 2025
Generated synthetic EMG to improve intent detection for wearable robotics control.
Conference on Neural Information Processing Systems (NeurIPS), Creative AI Track, 2025
Learned tidiness from demonstrations to arrange objects into orderly layouts.
International Conference on Intelligent Robots and Systems (IROS), 2024
Used scene graphs and LLMs to plan sub-tasks from instructions across diverse scenes.
Australasian Journal of Combinatorics, 2023
Defined and analyzed mathematical properties of the infinite Latin square.