Dr Amir Ghalamzan Introductory Seminar: Data-Driven Control for Effective Robot Physical Interaction

Title
"Data-Driven Control for Effective Robot Physical Interaction"

Abstract
In this talk, as the new Theme Lead for Robotics and Autonomous Systems at CMI, I will present my team’s recent work on data-driven predictive control for robotic grasp regulation. This research integrates real-time multi-modal sensor data with data-driven models to improve robotic manipulation and tackle challenges like slip prediction and grasp robustness. The core methodology involves learning world models from datasets of physical robot interactions, then embedding these models into predictive controllers to modulate reference trajectories and avoid slip during manipulation tasks. Experimental results show that this data-driven trajectory-based approach can outperform conventional grip force control in specific pick-and-place scenarios. I will also share my vision for Sheffield Robotics, emphasising the importance of interdisciplinary collaboration, responsible innovation, and impactful translation across manufacturing, healthcare, and society.

Short Biography
Dr Amir Ghalamzan is a Senior Lecturer (Associate Professor) at the University of Sheffield, serving as Director of Sheffield Robotics and Theme Lead for Robotics and Autonomous Systems at the Centre for Machine Intelligence. He heads the Intelligent Manipulation Lab, which specialises in robot learning, grasping and manipulation, teleoperation, agri-food robotics, and advanced tactile sensing. Amir received his PhD in Robotics and Automation Engineering from Politecnico di Milano, and has held previous academic posts at the University of Surrey and the University of Lincoln. His research prioritises data-driven control and planning methods for intelligent robotic systems and aims to solve practical problems in domains such as agri-food and healthcare, including robots operating in challenging, real-world environments.

Please register to attend