Max Robotics — Global Service Robot Directory + US Certification

About

Mingqiang Wei is a leading researcher in computer vision and 3D geometric processing, with a focus on scene understanding, object pose estimation, and point cloud analysis. His work bridges the gap between raw 3D data and high-level semantic interpretation, with over 89 citations across his most influential papers. Wei made significant contributions to RGB-D video segmentation by developing methods that robustly handle object detection and tracking under occlusion, enabling globally consistent segmentation across long video sequences—a critical advancement for robotic grasping and scene understanding. He pioneered multiscale feature line extraction from raw point clouds using local surface variation and anisotropic contraction, providing a way to extract meaningful structural cues from unstructured 3D data. His research on modeling indoor scenes with repetitions from raw point data has advanced automatic 3D reconstruction. More recently, Wei has been at the forefront of equivariant deep learning for 6D object pose estimation, developing SO(3)- and SE(3)-equivariant frameworks that leverage both RGB and depth information for robust object manipulation in robotics. His work on canonical shape reconstruction with SE(3) equivariance for weakly-supervised pose estimation represents a significant step toward more practical and data-efficient robotic perception systems.