Tool compensation in walk-through programming for admittance-controlled robots

This paper describes a walk-through programming technique, based on admittance control and tool dynamics compensation, to ease and simplify the process of trajectory learning in common industrial setups. In the walk-through programming, the human operator grabs the tool attached at the robot end-effector and “walks” the robot through the desired positions. During the teaching phase, the robot records the positions and then it will be able to interpolate them to reproduce the trajectory back. In the proposed control architecture, the admittance control allows to provide a compliant behavior during the interaction between the human operator and the robot end-effector, while the algorithm of compensation of the tool dynamics allows to directly use the real tool in the teaching phase. In this way, the setup used for the teaching can directly be the one used for performing the reproduction task. Experiments have been performed to validate the proposed control architecture and a pick and place example has been implemented to show a possible application in the industrial field.