Efficient camera-space manipulation using moments

摘要

The paper introduces a new estimation approach to the method of camera-space manipulation-a robust and precise means of controlling 3-dimensional robot maneuvers using vision. This approach reduces the computational and memory burden required by the formerly used "batch" estimation approach while retaining identical results. The discussion includes actual experimental results wherein reliable, millimeter positioning precision with a 3-dimensional, rigid-body task is achieved using a very large GMF S400 robot.