A virtual force sensor for interaction tasks with conventional industrial robots

The attempt to use industrial robots for technological and interaction tasks, i.e., robotic machining and robotic assembling, implies on the one hand the knowledge of the interaction force, on the other hand the reduction of physical sensors. The aim of this work is the development of a virtual force sensor to estimate the interaction force between a conventional industrial robot and the environment. The goal is achieved by exploiting a task oriented dynamics model calibration combined with of a thermal friction model of the robot. The dynamics model is calibrated by means of exciting trajectories made by suitable paths selected by a genetic-based two-stage optimization. The virtual sensor is proven by means of a polishing application. The proposed approach is successfully compared with state-of-the-art approaches. Finally, the use of the virtual force sensor in a closed-loop architecture highlights the effectiveness of the method in real applications.