Papers
8
Total Citations
413
H-Index
7
About
Wei-Chiu Ma is a leading researcher at the intersection of computer vision, robotics, and graphics, whose work is shaping how machines perceive and interact with dynamic 3D worlds. His primary research areas include 3D scene understanding, human and object modeling, and predictive behavior forecasting. Ma’s most influential contribution is his pioneering work on forecasting pedestrian dynamics using game theory and deep learning, which has garnered over 157 citations and is critical for safe autonomous navigation. He also developed the highly-cited Deep Rigid Instance Scene Flow framework (133 citations), which enables precise motion estimation in self-driving scenarios by leveraging strong geometric priors. In 3D human modeling, Ma introduced S³ (Neural Shape, Skeleton, and Skinning Fields), a novel approach for constructing and animating realistic virtual humans, and has advanced neural implicit modeling for reconstructing vehicles and objects from sparse, real-world data. His recent work on MIRA explores mental imagery for robotic affordances, pushing the boundaries of counterfactual reasoning in manipulation tasks. With a growing citation impact exceeding 400, Ma’s research is foundational for next-generation autonomous systems and virtual environments.
Research Focus
Key Achievements
Top Papers
- 1Forecasting Interactive Dynamics of Pedestrians with Fictitious Play157 citations · 2017
- 2Deep Rigid Instance Scene Flow133 citations · 2019
- 3
- 4Mending Neural Implicit Modeling for 3D Vehicle Reconstruction in the Wild24 citations · 2022
- 5MIRA: Mental Imagery for Robotic Affordances13 citations · 2022
- 6Secrets of 3D Implicit Object Shape Reconstruction in the Wild.8 citations · 2021
- 7Deep Rigid Instance Scene Flow7 citations · 2019
- 8S3: Neural Shape, Skeleton, and Skinning Fields for 3D Human Modeling3 citations · 2021