About Me
I am a senior humanoid control engineer at Tesla, working on Optimus robot. Before moving to the Bay Area, I am a PhD student at Carnegie Mellon University (CMU). I worked in the Robot Exploration Lab and the Biorobotics Lab in the Robotics Institude. My research advisors are Professor Zachary Manchester and Professor Howie Choset. Currently, I am working on state estimation for legged robots and complex robotics systems.
I obtained my B.Eng in Computer Engineering and M.Phil in Electronic and Computer Engineering from Hong Kong University of Science & Technology (HKUST) in 2012 and 2015. During my undergraduate study I was the team leader of HKUST Robotics team for three years. This team represented Hong Kong to compete in ABU Robocon. During my graduate study I was the team leader of HKUST IARC team in 2014. The team performed very well in IARC competition, an aerial vehicle competition that has more than 20 years history. Some of my master research work contributed to the flight control algorithm on DJI product autopilot A3.
Before joining the CMU, I worked at DJI for 5 years. It’s my honor to work with this company to change the world with drone technology. Starting from algorithm engineer, I became as one of project technical directors leading the research and development of many famous drone and robotics products. Projects that I directly involved are: Inspire 1, Phantom 3, Phantom 4, Matrice 100, Mavic, and various robotics products. The research drone platform Matrice 100 developed by me and my team is one of the most popular drone research platforms in recent years.
Other than robotics research, I am also enthusiastic about robotics education. At DJI I led an educational project called RoboMaster. In this project, we create different opportunities for students to work on practical robots and be recognized by the entire society. Many famous international medias have covered my work on these projects.
Research Statement
I use numerical optimization methods to enable legged robots to achieve stable control performance and precise low-drift long term state estimation. A key insight driving my research is control and estimation have inherent connections. The fact that control and estimation are dual problems allows mathematical tools developed for one problem to be applied to the other, such as factor graph and constrained trajectory optimization.
All of my research ideas must be proved by rigorous mathematical derivations or hardware experiments. All of my control and estimation software projects are opensourced on Github.
Robot Gallery
