Relative measurement for kinematic calibration using digital image processing

A camera-based robot measurement system is developed for the purpose of extracting relative measurements from sets of digital images for use in kinematic calibration. The system consists of a camera, lens, LED array, measurement head, data acquisition equipment and a precision-machined measurement artifact. A set of design requirements are established in order to drive the design of an image processing algorithm. The algorithm is developed using the MATLAB environment to process the digital images and extract metric information regarding the displacement error of a robotic end-effector in the X-Y plane of the robot workspace. Two novel post-processing algorithms are applied to the metric data in order to produce the relative measurements. One of these algorithms facilitates the removal of perspective distortion from image data, while the other is applied during cases where perspective distortion removal is not possible. The resolution provided by this relative measurement system is improved through the study of several parameters within the image processing algorithm and the final algorithm is characterized. This system is able to extract relative measurements to a mean resolution of +/-0.008 mm over a maximum displacement of +/-0.8 mm in the X direction and +/-0.5 mm in the Y direction. This mean resolution is approximately an order of magnitude less than the repeatability of the robotic manipulator to which the system is applied. Several suggestions for improving the resolution of this system are presented.