Robot learning from few demonstrations by exploiting the structure and geometry of data

Seminar | IST Alameda - DEI Informática II, room 0.19 | 11:00

Sylvain Calinon ,

EPFL – École Polytechnique Fédérale de Lausanne

Abstract:

Many human-centered robot applications would benefit from the development of robots that could acquire new movements and skills from human demonstration, and that could reproduce these movements in new situations. From a machine learning perspective, the challenge is to acquire skills from only few interactions with strong generalization demands. It requires the development of intuitive active learning interfaces to acquire meaningful demonstrations, the development of models that can exploit the structure and geometry of the acquired data in an efficient way, and the development of adaptive controllers that can exploit the learned task variations and coordination patterns. The developed models need to serve several purposes (recognition, prediction, generation), and be compatible with different learning strategies (imitation, emulation, exploration).

I will present an approach combining model predictive control, statistical learning and differential geometry to pursue such goal. I will illustrate the proposed approach with various applications, including robots that are close to us (human-robot collaboration, robot for dressing assistance), part of us (prosthetic hand control from tactile array data), or far from us (teleoperation of bimanual robot in deep water).

Bio

Dr Sylvain Calinon is a Senior Researcher at the Idiap Research Institute (http://idiap.ch). He is also a lecturer at the Ecole Polytechnique Federale de Lausanne (EPFL), and an external collaborator at the Department of Advanced Robotics (ADVR), Italian Institute of Technology (IIT). From 2009 to 2014, he was a Team Leader at ADVR, IIT. From 2007 to 2009, he was a Postdoc at the Learning Algorithms and Systems Laboratory, EPFL, where he obtained his PhD in 2007. He is the author of 100+ publications at the crossroad of robot learning, adaptive control and human-robot interaction, with recognition including Best Paper Awards in the journal of Intelligent Service Robotics (2017) and at IEEE Ro-Man’2007, as well as Best Paper Award Finalist at ICRA’2016, ICIRA’2015, IROS’2013 and Humanoids’2009. He currently serves as Associate Editor in IEEE Transactions on Robotics (T-RO), IEEE Robotics and Automation Letters (RA-L), Intelligent Service Robotics (Springer), and Frontiers in Robotics and AI.
Personal website: http://calinon.ch

For more information: