Experimental Evaluation of UR5e Collaborative Robot Force Control in Low-Force Applications

This article presents the findings of experimental research conducted to assess the stability of the force mode of the UR5e cobot from Universal Robots in the low-force range, from 1 N to 10 N. The set values of the robot's forces and the physically measured values were verified by an OptoForce Hex six-axis Force/Torque sensor attached to the robot's wrist, additionally coupled with an end-effector specially designed for research purposes. The results were recorded using proprietary software developed in the LabVIEW environment and a configured test lab station with a UR5e cobot. Three experimental tests were performed, in which the parameters of the effective force were measured while varying (1) the position of the task in the workspace of the robot, (2) the position and the level of force, and (3) the controller parameters of the force mode. The results of the experiments were compiled and presented in tables containing descriptions of, among other parameters, the following: the mean forces and their standard deviation; the mean maximum forces and its standard deviation; the mean root mean square error and its standard deviation; the mean absolute error and its standard deviation; the mean rate of force and its standard deviation; and the mean overshoot and its standard deviation. The findings of Experiment 1 demonstrated that when a setpoint of 10 N was employed, the UR5e cobot yielded an actual mean force ranging from 8.95 N to 13.26 N within the workspace plane. Experiment 2 showed that the average deviation from the set value within the 1-10 N range was approximately 0.38 N, with a maximum deviation of 0.61 N occurring at the limits of the working space. Experiment 3 showed that for the force range of 1-4 N, the best controller settings are Gain = 0.5 and Damping = 0.7; for the force range of 5-7 N: Gain = 1.0 and Damping = 0.6; and for the force range of 8-10 N: Gain = 2.0 and Damping = 0.8. Polynomial regression models were developed for each positioning scenario that can be used when making decisions regarding practical applications of the low-force mode.