Filling a glass of water: Continuously decoding the speed of 3D hand movements from EEG signals

We present a new system for the continuous decoding of hand movement speed in three-dimensional (3D) space from EEG signals. We recorded experimental data of five subjects during mimicking the natural task of filling a glass of water. The proposed system uses filter bank common spatial patterns and linear regression to estimate the speed of hand movements from artifact cleaned EEG signals. Average Pearson correlations between the speed trajectories predicted from EEG and the speed trajectories measured using a high-precision motion tracking system are r=0.41 for the x-axis, r=0.36 for the y-axis, r=0.48 for the z-axis, and r=0.17 for absolute speed in 3D space.