On the Applicability of Brain Reading for Predictive Human-Machine Interfaces in Robotics

• DOI: 10.1371/journal.pone.0081732
• Publication Date: 2013-12-16T21:27:04Z
• Authors: Elsa Andrea Kirchner; Su Kyoung Kim; Sirko Straube; Anett Seeland; Hendrik Wöhrle; Mario Michael Krell; Marc Tabie; Manfred Fahle
• Keywords: Robustness; Transfer of learning; Teleoperation

The ability of today's robots to autonomously support humans in their daily activities is still limited. To improve this, predictive human-machine interfaces (HMIs) can be applied better future interaction between human and machine. infer upcoming context-based behavior relevant brain states the have detected. This achieved by reading (BR), a passive approach for single trial EEG analysis that makes use supervised machine learning (ML) methods. In this work we propose BR able detect concrete interacting human. show detects patterns electroencephalogram (EEG) related event-related activity like P300, which are indicators or processes target recognition processes. Further, robustness applicability application-oriented scenarios identifying combining most training data classification applying classifier transfer. We testing, i.e., application classifier, carried out on different classes, if samples both classes miss pattern. Classifier transfer important usage scenarios, where only small amounts examples available. Finally, demonstrate dual an experimental setup requires similar as performed during teleoperation robotic arm. Here, movement preparation detected simultaneously. summary, our findings contribute development robust stable HMIs enable simultaneous behaviors.