Xianyi Cheng

Hi, I’m Xianyi (she-an-yee). I am an Assistant Professor at Duke Robotics. I obtained my PhD in Aug 2024 from Carnegie Mellon University, where I was fortunate to be advised by Matt Mason.

I’m interested in everything about robotic manipulation.
Manipulation is hard. Many manual tasks that are simple and repetitive for humans remain extremely difficult to automate.
Manipulation is intriguing. There’s still so much we don’t fully understand — from contact mechanics, the complexity of motions, controlling these complex motions, interactive perception, to problem solving for complex physical tasks. Manipulation is the intelligence of motion evolved over billions of years. Manipulation is the art of nature, now becoming the art through engineering and science.
Manipulation is the future. Human-level dexterity in robotics will transform industries and people’s lives. I aspire to help bring a future where robotic manipulation enables more people to pursue lives of dignity, purpose, creativity, and fulfillment.

email | google scholar | linkedin | twitter

Research

I’m proud to lead the Robot Dexterity Lab (DexLab), driven by amazing students full of passion and creativity. We focus on understanding and engineering manipulation. Currently, we are exploring questions such as:

How do we develop robot manipulation skills that are generalizable across tasks and transferable across robots?
What new forms of dexterity are possible, and can all robots be dexterous in manipulation?
What is the fundamental complexity of manipulation? Can we develop a formal framework to characterize its difficulty, robustness, and evaluation methods?
What infrastructure best supports the development, integration, and benchmarking of generalizable manipulation skills?

Teaching

Robotic Manipulation (ME-555), Spring 2025, Instructor, Duke
Dynamics (24-351), Fall 2022, TA and Weekly Recitation Instructor, CMU
Robot Dynamics and Analysis (24-760), Fall 2021, TA and Weekly Recitation Instructor, CMU