Motor prediction in Brain-Computer Interfaces for controlling mobile robots

EEG-based Brain-Computer Interface (BCI) can be regarded as a new channel tor motor control except that it does not involve muscles. Normal neuromuseular motor control has two fundamental components: (1) to control the body, and (2) to predict the consequences of the control command, which is called motor prediction. In this study, after training with a specially designed BCI paradigm based on motor imagery, two subjects learnt to predict the time course of some features of the EEG signals. It is shown that, with this newly-obtained motor prediction skill, subjects can use motor imagery of feet to directly control a mobile robot to avoid obstacles and reach a small target in a time-critical scenario. © 2008 IEEE.