Max Robotics — Global Service Robot Directory + US Certification

About

Xiangyang Ji is a prominent computer vision and robotics researcher whose work centers on 6D pose estimation, depth prediction, and intelligent robotic systems. He is perhaps best known for **DeepIM: Deep Iterative Matching for 6D Pose Estimation**, a landmark contribution that introduced a deep learning framework for iteratively refining object pose estimates — a paper that has accumulated nearly 800 citations across its versions, signaling its foundational influence on the field. Ji's research consistently bridges the gap between perception and action, tackling challenges such as class-level pose estimation for large object vocabularies using self-supervised learning, multi-camera depth prediction, and vision-based robotic assembly with RGB-only inputs. His group has also advanced point cloud processing through self-supervised implicit upsampling methods and contributed novel benchmark datasets for underwater robotics through the ROV6D dataset, addressing real-world deployment scenarios beyond controlled laboratory settings. More recently, Ji has explored efficient optical flow estimation and model predictive adaptation for robust robotic learning. Spanning autonomous driving, underwater robotics, and manipulation, his body of work reflects a sustained commitment to making 3D scene understanding practical, scalable, and deployable in complex real-world environments.