A review and performance comparison of visual servoing controls

Visual servoing is a technique for robot control, in which visual feedback is used in a closed-loop control to improve the accuracy and performance of robot systems. The control tasks in visual servoing are defined to control the robot using visual features extracted from the image. There are various problems when applying Visual servoing such as local minima, singularity, and visibility of feature points. These problems can be solved by using different features or using different control schemes. This paper provides a review of visual servoing for robot manipulators and conducts comparisons of five visual servoing approaches. First, the general theory of visual servoing and five different schemes for the comparison and evaluation are presented. Next, the behaviors of the Visual servoing system depending on the selection of visual features are also presented in detail. In addition, the enhancement and combination schemes are also presented, which are a combination of visual servoing with various control techniques that help increase the robustness of visual servoing. To overcome some issues of the image-based visual servoing scheme, different methods to approximate the interaction matrix are presented. After that, the five visual servoing schemes are simulated on Matlab for performance comparison and evaluation. To conduct the assessment, the simulations are implemented with typical control tasks that are translational movements and rotation around the X, Y and Z axes. The evaluations are conducted with the varying motion parameters and the varying effects of noise in the image. The results of the criteria are visually displayed as 3D charts, from which the reviews and comparisons of schemes are drawn. In addition, the paper also evaluates the schemes when performing general movements. General tasks are simulated by using the PUMA 560 robot. Each task has its own purpose to show issues in some schemes as well as how others overcome them.