Brain-machine interface based on deep learning to control asynchronously a lower-limb robotic exoskeleton: a case-of-study

This dataset includes the recordings that were used for a brain-machine interface (BMI) for controlling a robotic exoskeleton. EEG recording was carried out with 32 wet electrodes positioned over an actiCAP (Brain Products GmbH, Germany) at 500 Hz. Two additional electrodes, serving as ground and reference, were located on the ear lobes. Electrodes were placed following the 10-10 distribution, with four electrodes used for recording electrooculography (EOG), arranged in a cross shape with respect to the eye with the vertical ones around the left eye. The data were wirelessly transmitted using a WiFi MOVE unit (Brain Products GmbH, Germany) and amplified with BrainAmpDC (Brain Products GmbH, Germany).All files are MATLAB .mat and there are three types of files.Opened-loop filesRecordings obtained during opened-loop control, which is also referred as calibration in BMI. Participants performed 14 trials in four sessions with the exoskeleton in opened-loop control, during which they engaged in a series of mental practices including idle state and kinesthetic motor imagery. During half of the trials, participants stood still with the exoskeleton, and during the other half, they walked. The lower-limb exoskeleton was controlled the whole time by the predefined controlled periods.These files have the following name: A05_session4_openloop_4_motion.mat. First term refers to the subject, second term refers to the session, third term changes between opened- loop and closed-loop control trials, the following term is the index and finally, only for opened-loop trials, the last term identifies if the exoskeleton was moving during the trial or it was standing still.Each .mat file has two subfield: data_EEG and task_EEG. data_EEG is the EEG recording of all electrodes (electrodes x time) and task_EEG shows which mental task users were doing during each period. This value can range from 400 to 404. During 400 subjects did not have to think about anything in concrete. At the beginning of 401, subjects were given an acoustic cue that instructed them to start idle state period. This task has 4 seconds duration to capture evoked potentials. Then during 402, they continued being in an idle state. Afterwards, at the beginning of 403 they were also given an acoustic cue to start kinesthetic motor imagery period. This has also 4 seconds duration to capture evoked potentials. Finally, during 404 they continued performing motor imagery.NOTE: There were some issues with some of the trials. In A01 and A04, session 1, trial 14 were not properly saved, so they were corrupted.Closed-loop filesRecordings obtained during closed-loop control. Participants performed five trials in two sessions with the exoskeleton in closed-loop control. It means, that the movement of the exoskeleton depended on users’ thoughts.Each .mat file has three subfields: data_EEG, event_EEG and commands. data_EEG as in opened-loop files is the EEG recording of all electrodes (electrodes x time). event_EEG is a vector with the same size as data_EEG that shows in which part of the corridor subjects were at the moment. In this closed-loop control, participants had to go through a path in a corridor that was marked with color lines on the floor: yellow lines marked the beginning of the area in which subjects should walk and red lines marked the beginning of the area in which subjects should stop. These lines are identified in event_EEG as a 1 for red line and a 2 for yellow line. Finally, command shows the decisions made by the BMI for every 0.5 s. Since the sampling frequency is 500 Hz, this vector has the size of data_EEG/(0.5s * 500). A command value of 1 means that a WALKING command was sent to the exoskeleton, and a command value of 0 means that a STOP command was issued.Electrodes EEGThis file contains the position of the EEG electrodes. Therefore, if first electrode corresponds to F3, in the data_EEG matrix the first row will correspond to data from this electrode. The last four electrodes are EOG electrodes that were positioned forming a cross shape, so VU stands for vertical up, VD for vertical down, HR for horizontal right and HL for horizontal left.