An Intelligent Indirect-Hybrid Force/Position Controller for Smooth and Accurate Tracking of Unknown Contours

摘要

Various robotic tasks (e.g. polishing, deburring gluing, etc.) require a tool to be moved over some unknown surface with a constant force applied as well as keeping the orientation of the tool normal to the surface. While performing such fine motion control with a position-controlled industrial robot, it is required to sense the forces and accordingly manipulate the robot position and orientation. One such approach is Indirect Hybrid Force/Position Control that commands the robotic tool to be normal to the surface as well as the next position command so as to trace the surface, based on the forces experienced. In this paper, two improvements for the Indirect-Hybrid controller are devised. The first improvement, using fully-active sensing in place of semi-active sensing, leads to the better stability of the tool when the robot tries to make it normal to the surface. Another improvement replaces the unintelligent, non-predictive contour tracking with contour prediction using interpolation, regression and neural-network based techniques in the controller, leading to the accurate tracing of the surface. Finally the three predictive strategies are compared and the experimental results are demonstrated. The paper concludes by presenting an application for which the tracking algorithm is being used.