摘要: As the cost of sensors decreases and ability to model learn from multi-modal data increases, researchers are exploring how use unique qualities physically embodied robots help engage students promote learning. These designed emulate emotive, perceptual, empathic abilities human teachers, capable replicating some benefits one-on-one tutoring teachers. My thesis research focuses on developing methods for analyze integrate multimodal including speech, facial expressions, task performance build rich models user's knowledge preferences. student then used provide personalized educational experiences, such as optimal curricular sequencing, or leaning preferences style. In this abstract, we summarize past projects in area discuss applications learning affective signals transfer across tasks.