Industrial applications of camera space manipulation with structured light.

摘要

The dissertation presents theory and experimental validation of a generalizable approach for visual guidance of robots using a vision-guided robotic palletizing system, the first industrial application of Camera Space Manipulation (CSM) in day-to-day use, and a prototype of a robot vision system for truck metal-stamping assembly. Both applications utilize CSM and a multiple-laser-spot, image-difference-based illumination scheme to solve problems proven to be very difficult, if not impossible, for conventional methods.; A comparison is conducted between CSM and the mainstream methods of calibration and visual servoing. The intrinsic advantages for CSM are illustrated. Numerical simulations are conducted to demonstrate the robustness of the terminal precision of positioning to various errors in physical parameters which form the basis of the estimation model. These numerical simulations verify that CSM has little vulnerability to hardware errors, such as the errors in lenses and the robot forward kinematics. Therefore, CSM achieves robust, accurate and reliable positioning results using off-the-shelf hardware.