| Today, manufacturing robots have high repeatability but low accuracy. Manufacturers take advantage of their repeatability by using elaborate and expensive fixtures, together with robot path teaching. Our research aims at improving the accuracy of these robots, to eliminate this need for expensive fixtures and the time-consuming path teaching.;To improve the accuracy of manufacturing robots, we add sensors, e.g. laser tracker sensors, for real-time measurement of the positions of points on the robot's end-effector. Some practical limitations of these sensors are (1) they cannot measure orientation, and (2) they cannot be used to measure directly the position of the tool tip on the end-effector. Our research is directed toward generating real-time estimates of the orientation of the tool and of the position of the tool tip.;In this dissertation, we first present some experimental results on the error sources of a 6-axis gantry-style manufacturing robot. These error sources are geometric errors in the robot axes and temperature effects. These errors are the causes of low accuracy in today's manufacturing robots, which use no real-time measurements of the end-effector position or orientation. Then we describe the development of an adaptive estimation method to estimate, in real-time, this robot's end-effector position and orientation, using laser tracker measurements and all joint position measurements. We developed this adaptive estimation method to compensate for the geometric errors and the temperature effects. Finally, we present some experimental results demonstrating the performance of the adaptive estimation method. |
