Pingchuan Ma

Moscow Institute of Thermal Technology, Massachusetts Institute of Technology

Papers

Total Citations

H-Index

About

Pingchuan Ma is a robotics and artificial intelligence researcher whose work spans multi-objective reinforcement learning, computational robot design, and differentiable simulation. His most recognized contribution, "Prediction-Guided Multi-Objective Reinforcement Learning for Continuous Robot Control" (2020), has garnered 44 citations and introduced innovative methods for navigating complex trade-offs in continuous control tasks — a foundational challenge in deploying robots in real-world environments. Ma has also made meaningful strides in autonomous underwater vehicle (AUV) research, developing graph grammar-based computational frameworks for the automatic design of heterogeneous robot fleets, reducing the engineering burden traditionally associated with ocean-exploration robotics. His work on soft robotic fish further demonstrates his versatility, tackling the notoriously difficult sim-to-real gap by leveraging differentiable simulation to accurately model fluid-structure interactions and infer material parameters from real hardware data. Across these contributions, Ma demonstrates a consistent drive to bridge theoretical computation with physical robotic systems — whether navigating ocean floors or mastering soft-body mechanics — making him a distinctive voice at the intersection of machine learning, simulation, and embodied robotics.