Endeffector Force Approximation

Abstract

Endeffector forces play an important role in certain manufactoring tasks, such as screwing or inserting, in path tracking tasks, such as contour following or grinding, in tasks where e.g. two robots handle the same object and in mobile robots where the robot arm and consequently the end effector is influenced by platform movements. Before any force control scheme can be implemented there has of course to be a certain force measurement or sensing. In the literature there are mainly three approaches to force sensing. The first is the use of wrist force sensors, i.e. sensors consisting of mechanical elements equipped with strain gauges which convert displacement to force signals (see e.g. [SHIMANO, ROTH 79]). There are different designs available, but the principle remains the same. The advantage of these force sensors is the accuracy of the measurement and the fact that exact knowledge of the endeffector forces is obtained. Their disadvantage is the relatively high price, that makes them impossible to use when low cost solutions are requested. Another method is to measure or monitor the joint torques, which is either done by torque sensors mounted on each joint or by analyzing the actuator current [SHIMANO, ROTH 79], [NAGHDY et al. 85]. This was very popular for master-slave robots. By this method even forces that do not occur at the endeffector will be recognized and therefore situations where the arm hits obstacles will be realized. However there remains the costs for torque sensors.