Marco Capotondi

Papers

Total Citations

H-Index

About

Marco Capotondi is a researcher at the forefront of robotic control and learning, specializing in the challenging domain of underactuated and uncertain dynamic systems. His work masterfully bridges model-based control theory with data-driven learning, addressing the critical gap between idealized robot models and real-world physical interactions. Capotondi’s major contribution lies in developing online learning procedures that enable robots to adapt and refine their control policies in real-time, even without direct torque measurements. His most cited work (2021, 17 citations) introduces an iterative approach that estimates perturbations from model uncertainty on both active and passive degrees of freedom, allowing for robust planning and control of underactuated robots. This is complemented by his 2019 study (8 citations) on feedback linearization control, which demonstrates how a-priori dynamic estimates can be corrected online to achieve precise torque commands despite modeling errors. More recently, Capotondi has extended his expertise into robotic grasping (2022), exploring transfer and continual supervised learning through grasping features. His research is particularly notable for its practical impact on real-world robotic applications, offering scalable solutions for robots operating in unstructured environments where perfect models are impossible.