Max Robotics — Global Service Robot Directory + US Certification

About

Ryan Razani is a computer vision and machine learning researcher whose work centers on 3D scene understanding for autonomous driving and robotic systems. His research has made significant contributions to LiDAR-based semantic segmentation — the critical task of classifying every point in a 3D point cloud to enable safe, intelligent navigation in complex environments. Razani is perhaps best known for developing **(AF)²-S3Net**, an attentive feature fusion network with adaptive feature selection for sparse semantic segmentation, which has accumulated over 244 citations and stands as one of the most impactful recent advances in the field. Building on this foundation, he introduced **S3Net**, its architectural predecessor, alongside **TORNADO-Net** — a multi-view neural network incorporating total variation and a novel Diamond Inception module — and **Lite-HDSeg**, a lightweight approach leveraging harmonic dense convolutions for efficient LiDAR segmentation. Together, these works have garnered nearly 500 citations, demonstrating their broad influence across both academia and industry. Razani's research addresses a fundamental challenge in autonomous systems: enabling vehicles to perceive and interpret their surroundings accurately and efficiently in real time. His suite of architectures reflects a consistent drive to balance segmentation accuracy with computational practicality, making his contributions highly relevant to next-generation self-driving technologies.