Predictive Control Strategies for Image Based Visual Servoing of 6 DOF Industrial Robot

Abstract

The need for robotic process automation is increasingly being felt in the newlinemanufacturing industries. Presently, robotic systems are limited to operate in highly newlineunstructured environments. The integration of a vision sensor with the robot provides a newlinebetter perspective on their operations and aides the device to navigate the landscape and newlineavoid collisions. Visual servoing is a well-advanced method of controlling the robot newlinemovement based on visual sensor feedback. Based on the way the error is calculated to newlinegenerate the velocity profile, the aforesaid method is mainly categorized into two: newlineimage-based visual servoing (IBVS) and position-based visual servoing (PBVS). The newlineprojection of 3D information onto a 2D image plane in the camera obviously causes the newlineloss of data. This loss of data is a challenge in vision-based control. Moreover, the newlinenonlinearities and complex structure of a manipulator robot make the problem more newlinecomplex. newlineThe most primary controller used in visual servoing is a proportional controller that newlinereduces the image errors exponentially. Even if the proportional controller is easy to newlineimplement, its unsatisfactory behavior and the difficulty of constraint handling abstain newlineits application in industries. Model predictive controller (MPC) is adopted to solve newlinecomplex non-linear visual servoing control problem. IBVS has been expressed as a newlinenon-linear optimization problem in the image plane over a prediction horizon. Hence, newlinethe visibility constraints and velocity limitations of the robot are easily handled. The newlineproposed method gives convergence and satisfactory behaviors for complex newlinemovements. However, the method requires excessive computational time to solve the newlinequadratic function optimization. Hence, an adaptive horizon MPC is introduced in newlinewhich the horizon length of MPC is adjusted online as and when it is needed. In