Parallel Kinematics Manipulators for Pick and Place of Moving Objects Using 3D Perception: Development of a Test Bench

Interest in increasing the efficiency of manufacturing processes has led to a wide research effort focused on pick-and-place of moving objects. This topic presents a number of difficulties and consequently areas for research due to the need to integrate and optimize both the visual perception system, the image processing algorithm, and the robot control system. In the literature, many experimental efforts regarding new proposals to improve this process are carried out using anthropomorphic manipulators. The authors’ interest is instead to develop a testbed in order to investigate the various issues concerning dynamic pick-and-place, using a parallel kinematics manipulator. In fact, parallel kinematics manipulators are widely used in industry for high-speed pick-and-place tasks due to their inherent stiffness and dynamic performance. In this paper, a 4-degree-of-freedom 5R parallel kinematics robot will be integrated with a 3D camera for tracking moving objects. The manipulator and 3D camera will be presented, and then the software and hardware architecture of the testbed, resulting from the integration of the two systems, will be described. The object tracking algorithm used will be described, and analysis regarding its computational performance will be shown. Finally, considerations about the feasibility of integrating the two systems for dynamic pick-and-place will be drawn, deriving the future activities that are intended to be done.