Robotic measurement system: Self-calibration, real-time error compensation and path planning

This dissertation addressed three critical issues in developing a robot measurement system are addressed, i.e., system self-calibration, system accuracy maintenance and path planning. A self-calibration method, based on relative measurement instead of Cartesian coordinate measurement, is proposed to calibrate the robot measurement system on the manufacturing floor. Compared with traditional calibration methods this self-calibration is more advantageous in terms its implementation and the accuracy of parameter estimation. In order to achieve the high accuracy required by a robot measurement system, nongeometric errors such as compliance and thermally-induced errors have to be taken into account. The effect of compliance errors and thermally-induced errors on the robot measurement system performance is under investigation. A general methodology has been developed to calibrate these errors. Principal component analysis and orthogonal regression are used to construct the empirical models so that the thermally-induced error can be compensated for by monitoring the system's temperature field, therefore the system's accuracy can be maintained. An optimal path planning algorithm is developed for the robot measurement system with an area sensor. The method integrated techniques from geometric modeling, integer programming, principal component analysis and sphere package to generate the position and orientation for the sensor frame based on the object's geometric representation. Linear programming techniques has been used to determine the sequence so that the sensor follows a shortest path.