Robotic Contour Tracing with High-Speed Vision and Force-Torque Sensing based on Dynamic Compensation Scheme

Abstract

This research aims to realize robotic planar-contour tracing of unknown object with high-speed and high accuracy. The challenge lies in the existence of uncertainties, which can be attributed to a robot itself (such as modeling errors, mechanical defects like backlash) or due to environmental issues (like calibration errors, pose fluctuation of a workpiece). We propose to realize the tracing task by a non-model-based dynamic compensation approach based on the coarse-to-fine philosophy, which enables the manipulation with high speed and good accuracy simultaneously. This is achieved by adopting a methodology where a main robot is performing fast but coarse motion, while an add-on module is conducting accurate compensation of the overall uncertainties using high-speed visual feedback as well as force-torque feedback. Experiments for verifying the proposed method are conducted.