The Auctus team is delighted to receive the visit of Philip Long on November 6 and 7.
Dr. Philip Long is a Lecturer in Robotics & Automation at Atlantic Technological University, Galway (ATU), Ireland and currently PI of the SFI Robomate project. His research interests include human-robot collaboration, sensor-based control of robot manipulators, cable-driven parallel robots and flexible manufacturing. He has extensive experience in technology transfer projects in robotics, having worked in both industry and academia for over 15 years. Prior to joining ATU, at IMR, he was PI/Co-PI on a number of large-scale technology transfer and equipment projects in robotics totalling over €3M, including the Smart Eureka MAAS project, the Horizon 2020 projects CISC and ACROBA, and a DTIF award. During his time as a research engineer at IRT Jules Verne, Nantes, France, he developed sensor-based control schemes for human robot collaboration in partnership with industrial partners notably Renault, Airbus and Alstom (GE). Furthermore, he has contributed to high TRL university led projects developing multi arms robotic system for muscle separation and a U.S Department of Energy led project developing control architectures to enable a humanoid robot (NASA’s Valkyrie robot) to execute nuclear decommissioning. He holds a Mechanical Engineering Degree for University of Galway, a European Masters degree from University Genova and Ecole Centrale de Nantes, and a PhD from Ecole Centrale de Nantes. He has publications in several top tier robotics journals and international conferences, 4 international patents and is an associate editor for IEEE’s flagship robotics and automation conference ICRA- (International Conference for Robotics and Automation).
Abstract
To walk, run, jump or manipulate objects, robots need to constantly interact with objects and the environment. Unfortunately, reasoning about physical interactions is a computationally daunting task. For this reason, robots try to avoid physical interactions at all costs and unexpected physical contacts often lead to failures. In this talk, I will present our approach(es) to break down this complexity: the formulation of optimal control problems that leverage machine learning and numerical optimization to achieve real-time efficiency and real-robot robustness. I will also demonstrate our algorithms on real manipulation and locomotion examples. Finally, I will discuss current challenges towards real applications.