Aerospace Controls Laboratory

Google Scholar
macshen [at] mit [dot] edu
31-238A

Education

• Ph.D. candidate in Mechanical Engineering, MIT, 2017 - current
• M.S.E. in Ocean Engineering, University of Michigan, Ann Arbor, 2016
• B.S. in Ocean Engineering, Shanghai Jiao Tong University, 2015

Research Interests

• Deep reinforcement learning
• Multi-agent asymmetric adversarial games
• Autonomous vehicle

Projects

Context-aware Active Perception in Multi-agent Asymmetric Information Games
Macheng Shen, 2019

We are developing a general purpose algorithm for solving asymmetric information games that combines self-play for policy discovery, online POMDP planner for anticipation of opponent strategies, and maximum entropy agent modeling for robust belief update.