Task-level imitation learning using variance-based movement optimization | IEEE Conference Publication | IEEE Xplore