This dataset contains physical measurements of a human operator teleoperating the humanoid robot iCub. The operator controlled the robot while performing different tasks: reaching a low/mid-height target with the right hand, reaching a high target with the right hand, picking up a box located at different locations and placing the box at different locations. The operator's whole-body kinematics was recorded with inertial motion capture systems. All data files are in csv format. Training and validation data were used to learn movement primitives for the whole-body trajectories and were acquired when teleoperating the robot in a local network with ideally no delay. Compensated data show the whole-body trajectories used to control the robot in non-ideal networks affected by delays. A detailed description of how the data were collected and/or generated is available in the paper associated with the dataset: "Prescient teleoperation of humanoid robots" (Penco et al., hal preprint) https://hal.archives-ouvertes.fr/hal-?/document