Zelin Ye

Papers

Total Citations

H-Index

About

Zelin Ye is a robotics researcher whose work sits at the intersection of computer vision, reinforcement learning, and robotic manipulation. His primary research focuses on developing intelligent grasping systems that can operate effectively in cluttered, real-world environments—a notoriously difficult challenge due to partial occlusion and limited sensor feedback. Ye’s most cited work, “Transferable Active Grasping and Real Embodied Dataset” (2020, 19 citations), introduces a novel framework that combines reinforcement learning with 3D vision architectures to enable a robot to actively search for optimal grasping viewpoints using a hand-mounted RGB-D camera. This approach allows the system to transfer learned grasping skills across different scenes and objects, significantly improving robustness in unstructured settings. In his earlier work, “TendencyRL” (2019), Ye tackled the problem of sparse rewards in multi-stage robotic tasks by proposing a curriculum learning method that provides discriminative hints to guide the agent toward goal completion. Together, these contributions demonstrate Ye’s commitment to bridging the gap between simulated RL environments and real-world robotic applications, making his research highly relevant for students and engineers working on embodied AI and autonomous manipulation.