An EEG Adaptive Information System for an Empathic Robot

This article introduces a speech-driven information system for a humanoid robot that is able to adapt its information presentation strategy according to brain patterns of its user. Brain patterns are classified from electroencephalographic (EEG) signals and correspond to situations of low and high mental workload. The robot dynamically selects the information presentation style that best matches the detected patterns. The resulting end-to-end system consisting of recognition and adaptation components is tested in an evaluation study with 20 participants. We achieve a mean recognition rate of 83.5% for discrimination between low and high mental workload. Furthermore, we compare the dynamic adaptation strategy with two static presentation strategies. The evaluation results show that the adaptation of the presentation strategy according to workload improves over the static presenThis work has been supported in part by the Deutsche Forschungsgemeinschaft (DFG) within the Collaborative Research Center 588 “Humanoid Robots Learning and Cooperating Multimodal Robots” Dominic Heger · Felix Putze · Tanja Schultz Karlsruhe Institute of Technology Institute for Anthropomatics Cognitive Systems Lab (CSL) Adenauerring 4 D-76131 Karlsruhe, Germany http://csl.anthropomatik.kit.edu Dominic Heger Tel.: +49-721-608-46293 E-mail: [email protected] Felix Putze Tel.: +49-721-608-44328 E-mail: [email protected] Tanja Schultz Tel.: +49-721-608-46300 E-mail: [email protected] tation strategy in both, information correctness and completeness. In addition, the adaptive strategy is favored over the static strategy as user satisfaction improves significantly. This paper presents the first systematic analysis of a real-time EEG-adaptive end-toend information system for a humanoid robot. The achieved evaluation results indicate its great potential for empathic human-robot interaction.