Mediating between Qualitative and Quantitative Representations for Task-Orientated Human-Robot Interaction