Idiap Research Institute

Switzerland

The expertise of this partner includes human-centric robot applications with developing probabilistic approaches to encode movements and behaviours for robots evolving in human environments; model development (for recognition, prediction, online synthesis) shared by different learning strategies (imitation, exploration or incremental refinement); facilitating the transfer of skills from humans to robots, or in-between robots, by exploiting multimodal sensory information and by developing intuitive teaching interfaces; tools to 1) the representation and acquisition of manipulation skills as object-centric probabilistic models that can be combined with optimal control techniques, 2) the development of easy-to-use interfaces to transfer new manipulation skills to robots, as well as to visualize the learning progress of the robot. Its contribution to the project will involve the development of algorithms and representations to facilitate manipulation skills transfer and the proxy-tactile task-oriented control.