Over the past few decades, vision based alignment has been accepted as an important technique to achieve higher economic benefits for precision manufacturing and measurement applications. Also referred to as visual servoing, this technique basically applies the vision feedback information and drives the moving parts to the desired target location using some appropriate control laws. Although recently rapid development of advanced image processing algorithms and hardware have made this alignment process an easier task, some fundamental issues including inevitable system constraints and singularities, still remain as a challenging research topic for further investigation. This paper aims to develop a visual servoing method for automatic alignment system using model predictive control (MPC). The reason for using this optimal control for visual servoing design is because of its capability of handling constraints such as motor and image constraints in precision alignment systems. In particular, a microassembly system for peg and hole alignment application is adopted to illustrate the design process. The goal is to perform visual tracking of two image feature points based on a XYθ motor-stage system. From the viewpoint of MPC, this is an optimization problem that minimizes feature errors under given constraints. Therefore, a dynamic model consisting of camera parameters and motion stage dynamics is first derived to build the prediction model and set up the cost function. At each sample step the control command is obtained by solving a quadratic programming optimization problem. Finally, simulation results with comparison to a conventional image based visual servoing method demonstrate the effectiveness and potential use of this method.

No abstract

No abstract

This paper presents a visual feedback control system that is based on an efficient second-order minimization algorithm and can be used to track a moving face. This study aims to control a camera installed at the end of a three-degreeof-freedom robot arm and fix its orientation with respect to a moving face so that the face remains at a constant position in the image. In this work, the efficient second-order minimization algorithm is used to feed back the orientation of the two-dimensional human face. Then, an intelligent fuzzy algorithm is utilized by the tracking system to allow it to both track and become synchronized with the movement of the face. In addition, a Kalman filter is implemented to create an improved performance of the prediction of the position of the face. The applied control scheme is tested in tracking experiments on a moving face performed using a robot system with a camera installed on its end-effecter and the proposed scheme is shown to be an effective approach to the considered problem.