Integrating vision, touch and natural language in the control of a situation-oriented behavior-based humanoid robot

In the personal or service robotics domain very close interaction between humans and robots is crucial. To facilitate such a close interaction we propose to design the robot according to an anthropomorphic model, to combine visual and tactile sensing and to base human-robot communication on natural language. Furthermore, we argue that integrating these key technologies into a robot requires a certain kind of behavior-based system architecture that relies on an understanding of situations for the selection of the behavior to be executed. A robot implementing this architecture allows a human interface to be realized that makes the robot appear intelligent and easy to communicate and interact with. Results of first experiments are reported.