Learning Multimodal Confidence for Intention Recognition in Human-Robot Interaction