Abstract

The dataset is related to programming by demonstration and physical human–robot interaction, with a specific focus on the analysis of interaction forces generated at the robot end-effector during contact-rich task execution. It includes force measurements collected during the autonomous execution of two industrially relevant tasks—a robotic wiring task and a welding-like surface contact task—performed with and without compliance programming. The dataset supports the quantitative evaluation of the effects of variable impedance control on interaction behavior, enabling comparison between compliant and noncompliant executions in terms of peak contact forces, interaction stability, and task safety. The data were acquired from experimental trials in which users programmed both robot trajectories and interaction behaviors via kinesthetic teaching. During teaching, robot compliance was modulated by the users through a bio-inspired interface based on muscular cocontraction estimation, and the resulting compliance profiles were replayed during automatic task execution. Interaction forces at the robot end-effector were recorded during task execution to assess how compliance programming influences force escalation, friction-induced blocking, and overall interaction quality. The dataset enables statistical analysis of peak force values across subjects and task phases, as presented in the associated publication: R. Meattini, A. Govoni, K. Galassi, D. Chiaravalli, G. Palli, and C. Melchiorri, “Programming Robot Interaction Behavior During Kinesthetic Teaching Exploiting sEMG-Based Interfacing and Vibrotactile Feedback”, IEEE/ASME Transactions on Mechatronics, vol. 30, no. 5, pp. 4011–4022, 2025, DOI: 10.1109/TMECH.2025.3603402.