A formulation of automatic parameter tuning for robot arms

Fine tuning parameters of robots is a time-consuming and costly procedure. Many commercial robot companies can not afford to do fine tuning work for robots. Thus, the physical capability of robots can not be fully realized. This becomes a particularly serious problem when we want robots to operate with high accuracy performance at high speed. In this paper, a optimization problem is formulated and a computer package is designed to automatically tune parameters of feedback compensators for robot arms in order to circumvent time consuming and labor intensive manual tuning while maximizing the capability of robot arms. This auto-tuning by learning package makes it possible to accomplish the following: (1) fine tuning taking account of coupling effects between multiple-joints; (2) fine tuning for each robot; (3) retuning.