This paper presents a simple identification method of the actual kinematic parameters and compensation of each joint variable of articulated robot for improvement of positioning accuracy.
The positioning errors appear differently according to the teaching methods. This positioning error patterns due to the kinematic parameter errors were investigated. The kinematics of robot was refered. To model the kinematic parameter errors between consecutive near parallel joints by four parameter, the coordinate frames are reassigned.
The proposed identification method based on indirect measurement does not require a 3-dimensional measurement system and/or jigs but use only straight ruler about 1m long. A robot hand is posed using a teach pendant at the series of user specified points which are apart each other by an arbitrary distance of the ruler, and corresponding error functions are defined in terms of the kinematic parameters, encoder values and distance between teaching points. The identified kinematic parameters which get to zero of these error functions are obtained by iterative least square error method based on the singular value decomposition.
In the compensation of joint variables, to save computation time and implementate by on-line, the method of position-only compensation is developed, because the robot application to usual industrial processes, such as welding, painting and handling does not require a precise orientation control. Thus the orientation errors from the uncalibrated kinematic parameters were not considered in this compensation.
Finally, in the identification procedure, the effect of sampling errors of encoder value due to misteaching at point of ruler to the identified kinematic parameters is simulated to represent the sensitivity of the teaching error.
This method can be used directly at robot application line in workshop.