Datenbestand vom 30. November 2020
Tel: 089 / 66060798
Mo - Fr, 9 - 12 Uhr
Fax: 089 / 66060799
aktualisiert am 30. November 2020
978-3-8439-0631-9, Reihe Informatik
Towards Socially Interactive Robots – Designing an Emotion-based Control Architecture
220 Seiten, Dissertation Technische Universität Kaiserslautern (2012), Softcover, A5
Inspired by an ever older growing population in recent years the development of service and assistance systems for applications like elderly care, nursery, or entertainment and edutainment gained an enormous importance in the field of robotics. Compared to traditional application where robots cooperated with specially trained experts, these scenarios require a completely different set of abilities. The operation environment of these robots is the “every day life” of humans. Therefore, they need special interaction capabilities that enable socially acceptable behavior.
This thesis contributes to the development of flexible and adaptable control architectures for these socially interactive robots. The basic assumption is that the transfer of psychological concepts and ideas to a robot system leads to a control architecture that allows the robot to realize social interaction. On the basis of a thorough investigation of psychological theories on human’s socially interactive behavior core components have been identified. A concept for the realization of these components within a robot control architecture has been derived. To fulfill the requirements of flexibility and adaptability the different components provide uniform interfaces. Furthermore, guidelines for the generation of new components are given. This allows an integration of new components without requiring changes to the overall system.
For evaluating the derived design concept it has been prototypically implemented on the humanoid robot ROMAN. This robot is equipped with a complex sensor and actuation system including cameras and microphones to perceive its environment and over 40 degrees of freedom for generating interactive expressions. Various tests to demonstrate the function of the single subsystems have been realized. Furthermore, experiments on the quality of the robot’s expressions as well as on the robot’s interactive behavior have been conducted.
The design concept for building an emotion-based control architecture presented in this thesis can be used for further development in the area of social robots. The flexibility and adaptability that comes with the design allows for the integration of additional components that realize an automatic generation of new goals or the learning of new behaviors