Mingxuan Jing

Tsinghua University, Chinese Academy of Sciences

Papers

Total Citations

H-Index

About

Mingxuan Jing’s research lies at the intersection of robotic manipulation, skill acquisition, and reinforcement learning, with a focus on enabling robots to perform dexterous, adaptive tasks. A central contribution is their work on slip detection for stable grasping, where they pioneered a hybrid method combining unsupervised and supervised learning—using window matching pursuit for feature extraction—to improve robotic grip reliability during object handling. This foundational study has garnered 9 citations and directly supports safer, more autonomous manipulation in unstructured environments. Jing further advanced robotic learning through task transfer, introducing a preference-based cost learning framework that allows agents to migrate action policies between tasks without relying on explicit expert demonstrations or hand-coded costs—a significant step toward generalizable robot skill acquisition. Their work on robust tube-based model predictive control (MPC) for dexterous manipulation, published in 2024, offers smooth computational methods for high-precision control, while earlier research on dynamical movement primitives using wearable devices demonstrates a practical pipeline for transferring human skills to robots via teleoperation. With a growing citation footprint and contributions spanning perception, control, and learning, Jing is shaping the future of autonomous robotic systems.