Abstract
In this paper, two integrated target identification and acquisition algorithms and a graphical user interface (GUI) simulation tool for automated assembly of parallel manipulators are proposed. They seek to identify the target machine part from the workspace, obtain its location and pose parameters, and accomplish its assembling task while avoiding the collision with other items (obstacles). Fourier descriptors (FDs) and support vector machine (SVM) are adopted in this approach. The image of task area of workspace is obtained through machine vision, and the target assembling parts are identified. To acquire the location and pose information of the target, a modulus-shift matching (MSM) algorithm is proposed and integrated into the FD and SVM approaches, which could efficiently obtain the pose parameters while eliminating the effect of choice of starting point. The simulation results of two integrated algorithms, FD-MSM and SVM-MSM, are then compared and analyzed. In addition, a GUI is designed to visualize and assist the assembly process. An application on delta parallel robot with an extra rotational degree of freedom (DOF) is presented.
