Paul Guerrero

Adobe Systems (United States), University of Chile

Papers

Total Citations

H-Index

About

Paul Guerrero is a leading researcher in 3D computer vision and shape analysis, with a focus on unsupervised learning and geometric reasoning. His most influential work, "Unsupervised 3D Shape Reconstruction by Part Retrieval and Assembly" (2023, 11 citations), introduces a novel paradigm for representing 3D shapes as compositions of retrieved parts, enabling structural perception, robotic manipulation, and shape editing without requiring labeled data. This approach overcomes limitations of both parametric primitive methods and generative part spaces, offering a scalable, interpretable alternative for shape understanding. Earlier in his career, Guerrero contributed to multi-agent robotics with "An integrated multi-agent decision making framework for robot soccer" (2009, 3 citations), addressing decision-making challenges in the RoboCup domain, including ball placement and field positioning. His work bridges foundational robotics with cutting-edge 3D learning, demonstrating versatility from autonomous agents to geometric AI. Guerrero’s research is widely cited for its practical impact on shape compression, stylization, and robotic interaction, making him a key figure in advancing how machines perceive and reconstruct 3D environments.