DH Grollman, OC Jenkins - 2010 IEEE/RSJ International Conference on … - infona.pl
We propose to incrementally learn the segmentation of a demonstrated task into subtasks
and the individual subtask policies themselves simultaneously. Previous robot learning from …