Matej Urbas

Papers

Total Citations

H-Index

About

Matej Urbas is a researcher at the forefront of robotic manipulation and computer vision, with a focused expertise in enabling robots to handle deformable objects—a notoriously difficult challenge in automation. His most cited work, "Center Direction Network for Grasping Point Localization on Cloths" (2024, 12 citations), introduces CeDiRNet-3DoF, a deep-learning model that revolutionizes how robots identify and grasp non-rigid materials like fabrics. This contribution directly addresses the fundamental problem of object grasping in unstructured environments, where cloths’ variable shapes and textures defy traditional rigid-object approaches. By developing a network that predicts both center points and directional vectors for optimal grasping, Urbas provides a practical solution for applications in manufacturing, healthcare, and domestic robotics—such as laundry folding or surgical drape handling. His work stands out for its elegant fusion of geometric reasoning and neural network design, offering a scalable path for robots to interact with the everyday deformable world. With growing citation impact, Urbas is establishing himself as a key voice in robotic perception, pushing the boundaries of what autonomous systems can achieve with soft, unpredictable materials.