Including Uncertainty when Learning from Human Corrections

Dylan P. Losey, Marcia K. O'Malley

It is difficult for humans to efficiently teach robots how to correctly perform a task. One intuitive solution is for the robot to iteratively learn the human's preferences from co…