EMG and Motion Capture for 9\u2011Target Upper\u2011Limb Reaching in 22 Healthy Adults (10 EMG muscles, 9 markers)

This dataset contains synchronized surface electromyography (sEMG) and 3D motion\u2011capture recordings collected from 22 healthy right\u2011handed adults performing upper\u2011limb reaching toward nine spatial targets (ten repetitions per target; 90 reaches\/subject). sEMG was acquired from 10 muscles on the dominant side at 1000\u202fHz (biceps long, biceps short, brachioradialis, triceps, pectoralis clavicular head, deltoid anterior, deltoid medial, deltoid posterior, upper trapezius, teres major). Kinematics were captured with nine reflective markers at 250\u202fHz (neck, low back, clavicle, shoulder, external and internal elbow, external and internal wrist, hand. Kinematic data were interpolated with cubic spline when markers were occluded and were filtered with a low pass filter (3rd order Butterworth, cutoff frequency 3 Hz). Then they were upsamlped (4x) and synchronized with sEMG. Filtered and synchronized kinematic data (1000 Hz) were provided. Each subject file (MATLAB .mat) stores:(i) a structure \s\ with nine tasks (one per target), each containing:\u2022 \marker_filt\ (N\u00d727 double): filtered Cartesian coordinates (x,y,z) of the 9 markers in columns (order of markers is given in \info_marker_coordinates\;\u2022 \EMG raw\ (10\u00d7N double): raw EMG channels (each raw corresponds to a channel; order of muscle is given in \info_muscle_name\);where N is the total number of samples with sampling frequency equal to 1000.\u2022 \initial_window_EMG\ (scalar indexes):  onset segmentation point of each repetition (within the same task( derived from a manually selected minimum\u2011velocity threshold on the derivative of internal wrist marker (x-coordinate, antero\u2011posterior component);(ii) info_marker_coordinates (ordered cell array of marker names) and (iii) \info_muscle_name\ (muscle names per EMG channel). The dataset supports research on movement\u2011intention decoding, EMG feature design, time window\u2011length effects, and intra\u2011 vs inter\u2011subject validation for rehabilitation robotics and human\u2013machine interfaces. A JBHI manuscript (in submission) provides a systematic analysis of muscles, features, window lengths, and validation schemes. Please cite the dataset DOI and the paper when using these data. (Acquisition hardware, protocol, and filtering\/segmentation details follow the article\u2019s Methods.)