Abstract
An experiment was conducted to examine the utility of pressure and positional data from the Synaptics TouchPad to discriminating between baseline and problematic or frustrating user circumstances. Our special interest is in the problem of multimodal sensing and fusion of data streams from speech, haptics, and vision during acquisition of complex skills. In particular we are interested in finding optimal ways of combining raw sensory input to do user state sensing (e.g., emotions) and detect and forestall user errors or otherwise make appropriate adjustments during problematic situations, and also adjust system feedback to a level that best meets the needs of the individual user. This paper reports on our preliminary work on creating an emotion recognizer from the haptic data stream.