english only
EPFL > STI > IMT > LASA > Publications > Abstract



In robot Programming by Demonstration (PbD), the interaction with the human user is key to collecting good demonstrations, learning and finally achieving a good task execution. We therefore take a dual approach in analyzing demonstration data. First we automatically determine task constraints that can be used in the learning phase. Specifically we determine the frame of reference to be used in each part of the task, the important variables for each axis and a stiffness modulation factor. Additionally for bi-manual tasks we determine arm-dominance and spatial or force coordination patterns between the arms. Second we analyze human behavior during demonstration in order to determine how skilled the human user is and what kind of feedback is preferred during the learning interaction. We test this approach on complex tasks requiring sequences of actions, bi-manual or arm-hand coordination and contact on each end effector.


Downloadable files: 0) { $tempFile = $row['pdfFile']; $temp = "pdf"; echo "[$temp] "; } // ps.Z if (strlen($row['psZFile'])>0) { $tempFile = $row['psZFile']; $temp = "ps.Z"; echo "[$temp] "; } // ps.gz if (strlen($row['psgzFile'])>0) { $tempFile = $row['psgzFile']; $temp = "ps.gz"; echo "[$temp] "; } ?>



Last update: 25/08/06