A Fuzzy System for Detection of Interaction Demanding and Nodding Assent Based on Stereo Vision
Eugenio Aguirre, Miguel Garcia-Silvente, Antonio Gonzalez, Rui Paul, Rafael Munyoz
Abstract
Despite of the advances achieved in the past years
in order to design more natural interfaces between intelligent
systems and humans, there is still a great effort to be done.
Considering a robot as an intelligent system, the determination
of some typical interaction situations is an interesting ability to
achieve. This paper shows a fuzzy system that allows the visual
detection of possible interaction demands and the shaking or
nodding of the head. To achieve this objective, the robot has first
to carry out the detection and tracking of the people using the
stereo vision system. Then, the level of interest of a person to
interact with the robot is calculated by analyzing his/her position
and the pose of his/her head. The head pose is estimated in realtime
by a view based approach using Support Vector Machines
(SVM). Whenever the level of interest achieves an high value,
the person is analyzed in more detail to detect the position and
the motions of his/her arms as well as whether the person is
shaking or nodding his/her head. This information is managed
by a fuzzy system to detect a possible interest demand or the
intention of the person to say yes or no using his/her head. At
the end of the paper, some experiments are shown to validate
the proposal.Finally, future work is addressed.
in order to design more natural interfaces between intelligent
systems and humans, there is still a great effort to be done.
Considering a robot as an intelligent system, the determination
of some typical interaction situations is an interesting ability to
achieve. This paper shows a fuzzy system that allows the visual
detection of possible interaction demands and the shaking or
nodding of the head. To achieve this objective, the robot has first
to carry out the detection and tracking of the people using the
stereo vision system. Then, the level of interest of a person to
interact with the robot is calculated by analyzing his/her position
and the pose of his/her head. The head pose is estimated in realtime
by a view based approach using Support Vector Machines
(SVM). Whenever the level of interest achieves an high value,
the person is analyzed in more detail to detect the position and
the motions of his/her arms as well as whether the person is
shaking or nodding his/her head. This information is managed
by a fuzzy system to detect a possible interest demand or the
intention of the person to say yes or no using his/her head. At
the end of the paper, some experiments are shown to validate
the proposal.Finally, future work is addressed.
Full Text: PDF