|
|
|
|
|
ABSTRACT In recent years, there has been agreement regarding the notion of involving emotions in HRI, especially in a socially assistive robotics (SAR). On-line system, that recognized emotional state of the user and respond on it, could benefit effectiveness of SAR interaction. Emotional interaction is the multi-model system that consists of the face, voice/speech expression of emotion and, newly emerging aspect, BEE, which is the subject of the presented study. Our off-line vision-based recognition system recognizes BEE with relatively high recognition rate, taking as input still photos of subjects, picturing basic emotions (Anger, Fear, Sadness, Joy). Based on this algorithm an on-line system, for recognition of BEE and for expression bodily emotional states of the robot, could be developed. Then I will present the study described how BEE are perceived and interpreted. The eye movements of human subjects were recorded while they were observing and recognizing photographs of static body postures expressing basic emotions. We were interested in determining recognition accuracy, response times and eye fixation patterns. For the study human arm movement under influence of basic emotions we selected as of periodic movement the drawing of an ellipse. Both, geometrical and kinematical characteristics of movement have been analyzed, and the correlation between those characteristics (by power –laws) has been evaluated and shown that most of all characteristics are influenced by emotions. Summarizing the talk I will discuss how the presented studies may be applied in SAR. SPEAKER BIO Marina Fridin is completing her PhD study in Robotic Laboratory, Department of Applied Mathematics and Computer Science at Weizmann Institute of Science in Israel, advised by Professor Tamar Flash. Her doctoral work has been focused on Body Expression of Emotion for human computer-robot interaction and for motor control in humans and robotic systems. This research has been a part of the COBOL (COmmunication with Emotional BOdy Language) project, which broke new ground in advancing the nascent ability of machines to infer cognitive-affective mental states from nonverbal expressions of people. In the frame of her doctoral dissertation experimental work in human subjects has been combined with the development of mathematical and computational models that formulate hypotheses of gestures and motion during felling emotion. She presented her work on 6 scientific meetings of Europen Comission. Marina Fridin’s M.Sc. dissertation, glove-based gesture recognition and artificial gesture language, has been done also under supervising of Professor Tamar Flash, the Robotic Group at Weizmann Institute. The gesture-based language and an interface as an alternative to the traditional input interfaces have been developed and implemented. This project was done in collaboration with Technion University, inventing new technologies for needs of the Israeli Army (funded by MAFAT). At present time Marina Fridin is applying for post-doctoral position. Her scientific interests are in Human Computer Interaction, Robotics, Affective computing, Motor control, Machine learning, and Computer vision. Marina Fridin Email: marina.ousov-fridin@weizmann.ac.il |
|
Back to CRES |