Find out more about subscribing to add all events.
This talk presents a unified approach to estimation and system identification for contact-rich robots, where localization, inertial parameters, friction, disturbances, and contact interactions are estimated together rather than in separate stages. The key idea is to enforce the physics directly within the estimation problem, combining robot dynamics, contact behavior, and prior knowledge in a single structured optimization framework. By exploiting this structure, the method aims to achieve both physical consistency and computational efficiency, while providing a more realistic treatment of robot–environment interaction during complex motions.
Sergi Martínez received an MSc degree in automatic control and robotics from Universitat Politècnica de Catalunya (UPC), in 2022, and a BSc degree in mechatronics, industrial electronics and automatics engineering from the University of Vic (UVIC-UCC), in 2019. He is currently persuading a PhD at Heriot-Watt University under the supervision of Dr Carlos Mastalli. Previously, he worked as a Research Support Engineer at the Institute of Robotics and Industrial Informatics (IRI-CSIC), under the supervision of Guillem Alenyà. He was in charge of developing robotics applications for cloth manipulation in healthcare tasks. He has also been enrolled within the High Dynamics Group at IRI, where he conducted his master’s thesis on embedding actuators dynamics in the optimal control model of a UAM. His doctoral project focuses on combining predictive control techniques with deep learning to enable dynamically-balanced loco-manipulation on legged robots.