Max Robotics — Global Service Robot Directory + US Certification

About

Albert Pumarola is a researcher whose work sits at the intersection of computer vision, robotics, and deep learning, with a particular focus on human body understanding and human-robot interaction. His most influential contribution, "GanHand: Predicting Human Grasp Affordances in Multi-Object Scenes" (2020, 170 citations), broke new ground by tackling the largely unexplored challenge of predicting how a human would naturally grasp objects from a single RGB image — a capability with profound implications for robotics and augmented reality applications. This work stands as a landmark achievement in hand geometry estimation and affordance prediction. Beyond grasping, Pumarola has made meaningful contributions to human motion prediction, developing attention-based deep learning models that account for robot-human handover dynamics and context-aware motion forecasting. His earlier work on cognitive architectures for service robots demonstrated a consistent interest in enabling machines to interpret and respond to natural language commands within complex, real-world environments. Across his career, Pumarola has consistently pushed boundaries in making robots more perceptive and responsive to human behavior. With over 200 cumulative citations, his research has established him as a notable voice in embodied AI, human pose estimation, and intelligent robotic systems.