Title of Submission
Predicting Human Interpretations of Affect and Valence in a Social Robot
Computer Science, MS
Major Advisor Name
Type of Submission
9am - 10:30am - Judging Session #1
In our research, we seek to understand what emotion people interpret from a social robot’s behavior, and the positive or negative valence of that affect.
In our study, annotators labeled the Anki Cozmo robot’s pre-scripted behaviors using 16 possible emotion labels. Then, we generated novel behaviors and raters assigned scores to emotion valence pairs (e.g., sadness and joy). We compared our model’s predictions of valence between the affective pairs and compared the results to the human ratings.
We conclude that robot designers cannot assume people will perceive affect or valence as designed, and make several suggestions for future work.