Fusion of CSP-derived features and Time-Domain Parameters for mind robotics control

In this paper, fusion of Common Spatial Pattern (CSP) analysis derived features and Time-Domain Parameters (TDP) is investigated for two-class motion imagery classification: left hand movement and right hand movement imagery. This fusion method is aiming at improving the performance of Brain-Computer Interface (BCI) systems in translating human thoughts into command to control certain robots, e.g. rehabilitation robots. In the BCI system applied in this paper, a Human-Friendly Interactive Bear Robot is implemented, whose bear-like appearance and interactive actions can alleviate the frustration and anxiety on the part of the user in the experiment. Moreover, off-line experiments are conducted to demonstrate the efficiency of the fusion method using both well-known open-source data and data obtained by the BCI system mentioned before.