Plenary Talk 1: A perspective on robot programming for advanced sensor-based applications and human-robot interaction
Abstract: Robots are increasingly used for advanced applications in which they have to execute a variety of tasks in partially structured or even unstructured environments. Very often they have to operate in the presence of or even in close collaboration with humans. Hence, robot control programs should allow a high variation in the task specification. They also need to deal with a lot of real-time sensor information and comply with a lot of constraints (e.g. the robot’s own physical limits, its environment, interaction with humans, safety). In addition, robot programming should be fast and intuitive, so that robot users can easily deploy their robots for new tasks or in new environments.
The robotics research group at KU Leuven is a corelab of Flanders Make, Belgium. An important focus is to address the challenges mentioned above. We develop a new approach to program robots in a flexible way by formulating every robot task – or ‘robot skill’ – as a model-based constrained optimization problem. We investigate how this model-based approach can be tied up with learning from human demonstration to allow fast deployment of the robot skill for a new application (e.g. a new context, environment, robot platform). To support human-robot interaction, we develop methods to recognize the human intent and predict the motion of the human. Based on screw theory, we investigate how motions and forces are best represented in skill programming such that the skill has the highest generalizing capability towards new applications.
This talk gives an overview of the approaches and example applications.
Joris de Schutter
KU Leuven, Department of Mechanical Engineering, Belgium
Flanders Make, Belgium
View Google Scholar Profile
Short bio: Joris De Schutter received the M.Sc. degree in mechanical engineering from the Katholieke Universiteit Leuven (KU Leuven), Leuven, Belgium, the M.Sc. degree from the Massachusetts Institute of Technology, USA, and the Ph.D. degree in mechanical engineering from KU Leuven. Following work as a control systems engineer in industry, he became a lecturer in the Department of Mechanical Engineering, KU Leuven, where he is currently a full professor and chair of the department. He teaches courses on kinematics and dynamics of machinery, systems and control, robotics, and optimization. His research interests include sensor-based robot control and programming, human motion modeling, and human–robot interaction. He pioneered in robot force control and in constraint-based robot programming. In 2018 he obtained an ERC Advanced Grant focusing on generalizing human-demonstrated robot skills.