Brain Computer Interface for Gesture Control of a Social Robot: an Offline Study

Abstract

Brain computer interface (BCI) provides promising applications in neuroprosthesis and neurorehabilitation by controlling computers and robotic devices based on the patient's intentions. Here, we have developed a novel BCI platform that controls a personalized social robot using noninvasively acquired brain signals. Scalp electroencephalogram (EEG) signals are collected from a user in real-time during tasks of imaginary movements. The imagined body kinematics are decoded using a regression model to calculate the user-intended velocity. Then, the decoded kinematic information is mapped to control the gestures of a social robot. The platform here may be utilized as a human-robot-interaction framework by combining with neurofeedback mechanisms to enhance the cognitive capability of persons with dementia.

Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.