Hybrid eye-to-hand and eye-in-hand visual servoing for autonomous robotic manipulation

摘要

This paper presents a hybrid fuzzy control approach for a robotic manipulator to perform assembly tasks. Hybrid force and pose control typically employs an expensive high dimensional force sensor in addition to vision sensors. In contrast to traditional approaches, a low-cost single-axis force sensor together with CCD cameras and a laser projector can be applied to carry out hybrid force and pose control of a robotic manipulator on 3-D surfaces. Specifically, design of fuzzy control approaches has been applied to hybrid eye-in-hand and eye-to-hand visual servoing configurations. In particular, to expand the workspace while improving control accuracy, eye-to-hand approach is considered for target positioning in a work space followed by pose and force control employing eye-in-hand approach. Color features are designated as target points on a 3-D surface. After visually determining the direction normal to the 3-D surface at the center of the target through reconstructing the laser cross projected by the laser projector, a robotic manipulator equipped with a low-cost single-axis force sensor is able to perform hybrid force and pose control for drilling, bolting, and screw locking tasks. The proposed approaches have been successfully validated on 3-D surfaces by performing experiments with an industrial robotic manipulator.