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

We contribute a learning from demonstration approach for robots to acquire skills from multi-modal high-dimensional data. Both latent representations and associations of differ- ent modalities are proposed to be jointly learned through an adapted variational auto-encoder. The implementation and results are demonstrated in a robotic handwriting scenario, where the visual sensory input and the arm joint writing mo- tion are learned and coupled. We show the latent representa- tions successfully construct a task manifold for the observed sensor modalities. Moreover, the learned associations can be exploited to directly synthesize arm joint handwriting motion from an image input in an end-to-end manner. The advan- tages of learning associative latent encodings are further high- lighted with the examples of inferring upon incomplete input images. A comparison with alternative methods demonstrates the superiority of the present approach in these challenging tasks.

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