Michelle Guo
Papers
4
Total Citations
44
H-Index
2
About
Michelle Guo is a rising researcher at the intersection of robotics, computer vision, and physics simulation. Her work focuses on enabling robots to perceive, simulate, and physically interact with objects in the real world. She is best known for pioneering differentiable physics simulation for "neural objects"—3D representations like NeRFs that encode geometry as continuous density fields. Her 2023 paper on this topic, with 30 citations, introduces a pipeline that estimates dynamical properties directly from these learned fields, allowing robots to simulate and predict object motion in a fully differentiable manner. This breakthrough bridges the gap between visual perception and physical interaction. Guo also tackles the challenge of dexterous manipulation: her 2022 work combines generative models with bilevel optimization to learn diverse, physically feasible grasps for multi-fingered hands. Most recently, she explores an innovative direction—designing 3D-printable adaptations for everyday objects to make them easier for robots to manipulate, addressing the fundamental mismatch between human-centric design and robotic capabilities. Her research, though early-career, is already shaping how robots understand and act upon the physical world.
Research Focus
Key Achievements
Top Papers
- 1Differentiable Physics Simulation of Dynamics-Augmented Neural Objects30 citations · 2023
- 2
- 3
- 4Differentiable Physics Simulation of Dynamics-Augmented Neural Objects2 citations · 2022