Max Robotics — Global Service Robot Directory + US Certification

About

Qiyu Kang is a researcher specializing in 3D computer vision, autonomous systems, and deep learning, with a particular focus on perception and localization tasks critical to robotics and autonomous driving. Their work addresses some of the most challenging problems in spatial understanding, including point cloud registration, LiDAR-based pose estimation, and multi-modal place recognition. Kang's most influential contribution, "PointDifformer" (2024, 27 citations), introduces a novel integration of neural diffusion processes and transformer architectures to achieve robust point cloud registration under noisy and perturbed conditions — a persistent challenge in real-world 3D vision pipelines. Their work on "HypLiLoc" (2023, 19 citations) demonstrates a creative application of hyperbolic geometry to LiDAR pose regression, improving accuracy and computational efficiency in relocalization tasks where traditional database retrieval methods fall short. More recently, "PRFusion" (2024) advances multi-modal place recognition by fusing image and point cloud data to build more resilient and generalizable systems. Through these contributions, Kang has established a reputation for developing principled, geometry-aware deep learning solutions that push the boundaries of robust perception, making their work highly relevant to researchers and engineers working at the intersection of computer vision, robotics, and autonomous navigation.