A new method for closed-loop stability prediction in industrial robots

Abstract

With the demand for higher position accuracy from industrial robots used for precision manufacturing tasks, a common solution approach is to implement closed-loop feedback control using external sensors. Because most industrial robot controllers only allow real-time commands to be specified in the form of Cartesian or joint position offsets, the plant models of these closed-loop systems tend to be very simple in that they assume that the robot executes each input command with minimal or no error. However, real-time motion error can be of the order or larger than the corresponding input commands. Due to the shortcomings of these simplistic models, closed-loop controller gains need to inevitably be tuned manually through trial and error. If the missing components of the simplistic plant models can be identified, closed-loop controller gains can be readily determined efficiently through simulation. In this paper, robot controller delay and robot dynamics are identified as the key missing components, and a new data-driven method for capturing the robot dynamics and a model for closed-loop stability prediction are established. The new model-based method is experimentally evaluated on a six degree-of-freedom (6-DoF) industrial manipulator. It is confirmed that the new method can be used to determine via simulation robot controller gains that ensure closed-loop stability without the need for iterative trial and error experimental gain-tuning.