Geometry-based 6D-pose visual servoing system enabling accuracy improvements of industrial robots

摘要

Jointed arm robots inherit a vast number of beneficial features like flexibility of movement and repeatability in combination with low investment costs. However, one main drawback, which most often disqualifies these robots for tasks with high accuracy requirements, is their low absolute accuracy. Suitable methods to overcome this problem belong to the field of visual servoing. In this context a 6D-pose visual servoing system has been developed at FAPS institute, using off-the-shelf structured light projection sensor systems. Thereby, the workpiece geometry itself functions as reference object and hence no specialized markers are needed, as used in the state of the art. Additionally, the system contributes to a comprehensive compensation of cumulated geometrical errors, which are induced by the robot and the end-effector at the same time. As a byproduct of this added value, a method for determining the pose of virtual coordinate systems has been developed, allowing an accurate calibration and measurement of system components relative to each other. The effectiveness of the researched methods has been validated in course of experimental investigations, which show a good absolute accuracy of the 6D-pose correction in combination with a very high repeatability.