| High speed and high accuracy seam tracking for increased quality and productivity is limited by factors such as backlash, link flexibility, robot dynamics variations, servo characteristics, etc. An even more important limitation is imposed by the unavoidable mismatch between the actual trajectory to be followed and the programmed nominal trajectory caused by manufacturing and location errors of the workpiece.; Commercial computer vision and ultrasonic sensors are not only slow for fast moving robots but also expensive. Therefore, the aim of this research consisted in the development of a high speed and cost effective sensor system in conjunction with an adaptive controller for accurate seam tracking.; The design of the sensor system was based on the use of the image of the seam on a linear detector to extract position and orientation information about the seam. The three quantities extracted were the lateral deviation, distance variation, and angular deviation perpendicular to the direction of motion of the actual from the nominal seam. The achieved resolutions were 0.001in, 0.01in and 1{dollar}spcirc{dollar} respectively. A detailed mathematical model of the sensor operation along with a sensitivity analysis was introduced. With the linear array detector used to detect the seam location speeds of up to 1000 frames/sec could be achieved.; The controller proposed for seam tracking was based on the developed sensor and an adaptive joint controller designed around the pole placement principle. The unique feature of the controller scheme was the use of the predicted seam errors as corrections superimposed onto the nominal sensor commands.; The overall system was implemented in a multiprocessor environment capable of performing the image processing, joint dynamics identification, pole-assignment, and Cartesian to joint error transformations at sampling rates of 0.0055 Second.; It has been shown that the sensor combined with the adaptive controller could successfully follow the actual seam within close tolerance limits. |
