SENSEI - SENSING EMOTION NON-INVASIVELY FOR SOFTWARE-BASED EDUCATIONAL INSTRUCTION

DARPA and DOD often make use of intelligent tutoring systems (ITSs) for administering training and education to their employees. Unfortunately, even the most modern electronic tutors cannot capture the student"s emotional states and provide no adjustment accordingly. The most effective learning occurs when the instructor engages students according to their current emotional state. To this end, UtopiaCompression (UC) is developing a multi-modal affect recognition system that can estimate the subject emotional state using a number of non-invasive external cues. UC"s solution includes a set of innovative feature extraction and analysis modules for various sensory data and a novel probabilistic fusion model that incorporates complementary cues to aid estimation of the subject"s affective state efficiently and robustly. UC"s affect recognition technology will vastly improve the effectiveness of DoD"s educational programs through accurate, efficient and non-invasive sensing of the student"s emotions. It can also be used in vast other military and civil applications. Based on the successful Phase I feasibility study, UC will undertake the objectives enumerated in the Phase II proposal to develop a robust and reliable realtime affect recognition prototype system. The generated prototype system will be tested and demonstrated in a close-loop real ITS environment.