LIBRARY ı6ıul 0) - Cranfield University

3.1.1.1 Robot errors
A robot arm is a highly non-linear mechanical system which does not have a
truly closed-loop position feed-back control. The position and orientation of the end-
effector cannot be measured by normal sensors and it is usually estimated based on
non-linear inverse kinematics which assumes that the arm is perfectly constructed
according to the design specifications. In addition, the relative low stiffness of the
mechanical arm, allied with gear backlash, joint compliance, etc., may cause random
variation in positioning for repeated movements. This leads to errors between the
command and attained poses.
When using on-line programming in welding operations, a high repeatability
(see section 2.4.2) is very critical, since the robot merely plays back joint angles which
were previously recorded. Here the absolute accuracy is not relevant since
programmed points are set relative to the workpiece.
Tasks involving off-line programming, however, depend critically on the
absolute accuracy, further to the repeatability. A robot may have high repeatability
while having low absolute accuracy. Given the joint angles, the controller of a robot
calculates the pose of its end-effector with respect to a co-ordinate frame attached to
its base, based on a kinematic model of the manipulator structure. This model depends
on several parameters such as link lengths, joint offsets, joint compliance, gear
backlash, misalignment between parallel axes, link compliances, etc. Most of the time,
however, not all the parameters are taken into account or are accurately defined due
to manufacturing tolerances. Hence, the calculated poses will not match the required
ones, resulting in positioning errors. Therefore, robot errors originate mainly from a
robot's inability to achieve precisely the required co-ordinates, that is, from the lack
of absolute accuracy.
3.1.1.2 Programming errors
The programming errors have already been discussed in section 2.4.4. In
summary, programming errors in off -fine programming result mainly from the
mismatch between the ideal world and the real world, as defined in section 2.4.1.
Here, perfect kinematic models are used in both the simulation and the robot
controller to drive an imperfect arm. In addition, the graphical models of the
workpiece and cell environment are usually based on nominal dimensions, which are
subject to variations due to manufacturing tolerances.
3.1.1.3 Component errors
The component errors include: a) variation in joint shape (e. g. presence of gap
and joint misalignment); b) variation in part positioning due to inadequate fixturing
and/or presence of extraneous materials (e. g. spatter) on the fixture locating surfaces;
and c) variation in joint shape and position due to thermal distortion. Such errors are
the most difficult to deal with, since they vary from component to component within a
known tolerance range.
Another source of error that will be classified under this group is the deviation
of the wire tip from the torch axis due to the wire cast and contact-tip wear. This type
64