Andrea Rosasco

Papers

Total Citations

H-Index

About

Andrea Rosasco is a robotics researcher whose work lies at the intersection of 3D perception, action recognition, and human-robot interaction. Her key research areas include shape completion for robotic manipulation and skeleton-based action recognition, with a strong emphasis on enabling robots to operate robustly in unstructured, real-world environments. Rosasco’s major contributions address two critical challenges: first, she developed confidence-guided shape completion methods that allow robots to infer complete 3D object geometry from partial sensor data, overcoming the limitations of onboard visual sensors in cluttered workspaces. Second, she pioneered one-shot open-set skeleton-based action recognition, a framework that enables humanoid robots to learn new actions with minimal examples while reliably identifying and ignoring unknown actions—a vital capability for safe human-robot collaboration. Though early in her career, Rosasco’s work has already garnered citations (3 and 2 for her most-cited papers, respectively), reflecting its relevance to advancing practical robotic perception and learning. Her research stands out for addressing the real-world constraints of robotics—limited data, partial views, and the need for adaptability—making her a promising voice in the field of embodied AI and autonomous systems.