Ignasi Clavera

Papers

Total Citations

H-Index

About

Ignasi Clavera is a researcher advancing the frontiers of robotics and reinforcement learning, with a focus on enabling machines to operate in complex, unstructured environments. His work spans model-based reinforcement learning, non-prehensile manipulation, and robust perception systems. Clavera made a significant contribution with his 2017 paper on policy transfer via modularity and reward guiding, which tackled the notoriously difficult problem of robotic pushing—a task complicated by unknown frictional forces—by leveraging reinforcement learning to improve object manipulation without grasping. This work has garnered 39 citations, reflecting its impact on the field. He further advanced model-based reinforcement learning in his 2019 paper, proposing asynchronous methods that match the asymptotic performance of model-free algorithms while offering superior data efficiency, a critical achievement for real-world applications. More recently, in 2020, Clavera explored mutual information maximization to develop robust, plannable representations, addressing the challenge of high-dimensional state spaces in robotics. His research consistently pushes toward reducing sample complexity and enhancing system reliability, making him a notable figure in the quest for more capable, autonomous robots.