Static compensation for wrist mounted force torque sensor

This paper presents and evaluates a new technique for the calibration and compensation of static error of force/torque (F/T) sensors. The presented approach is based on a learning algorithm. A Modified Nearest Neighbor algorithm is presented to compensate non-linear errors, produced by the weight of the end effector or by the internal offset of a wrist mounted F/T sensor of a 6 degree of freedom (DOF) industrial robot.