New paper on active preference learning

Our results on Improving user specifications for robot behavior through active preference learning: Framework and evaluationhave appeared in the International Journal of Robotics Research (IJRR).  The preprint is available on arXiv.