ETRI-Knowledge Sharing Plaform

KOREAN
논문 검색
Type SCI
Year ~ Keyword

Detail

Conference Paper Adaptive Behavior Generation of Social Robots Based on User Behavior Recognition
Cited - time in scopus Share share facebook twitter linkedin kakaostory
Authors
Woo-Ri Ko, Minsu Jang, Jaeyeon Lee, Jaehong Kim
Issue Date
2020-09
Citation
International Symposium on Robot and Human Interactive Communication (RO-MAN) 2020, pp.1-4
Language
English
Type
Conference Paper
Abstract
Social robots should understand the user’s behavior and respond appropriately for natural human-robot interaction. Our work is aimed at recognizing the subtle differences in user behavior and generating behavior appropriately. For the user behavior recognition, we use a Kinect v.2 sensor for skeletal tracking and a deep neural network (DNN) for behavior classification. The weights of the DNN are trained using AIRAct2Act, which is a human-human interaction dataset. For the robot behavior generation, we designed several behavior selection rules by referring to the interaction scenarios of the dataset, and then modify the key pose of the selected behavior taking into account the user’s posture, position, and physical characteristics such as height. To demonstrate the effectiveness of the proposed method, we perform experiments using a Pepper robot in the 3D virtual environment. The experimental results show that the proposed method has a 98% accuracy in recognizing the user’s behavior and can naturally change the behavior in a situation where the user’s intention is confused.
KSP Keywords
3d virtual environment, Adaptive behavior, Behavior Generation, Behavior classification, Behavior selection, Deep neural network(DNN), Human-Human Interaction, Human-Robot Interaction(HRI), Natural human-robot interaction, Physical characteristics, User behavior