Adaptive Visual Servoing of Robotic Systems With Closed Structure and Limited Image-Space Position Measurement Capacity

Abstract

Existing torque-based visual servoing controllers are not applicable to industrial/commercial robots with closed structure. On the other hand, velocity-based visual servoing controllers operate effectively under the assumption that the torque controller can ensure the stability of the combination of torque control/visual control loop, while ignoring the effects of unknown dynamics. This article presents a new adaptive visual servoing loop controller, which dynamically generates joint position or velocity commands for the torque control loop based on image-space feedback. It is rigorously proven that the proposed scheme can guarantee the stability of the visual servoing system, despite uncertainties in the camera model, robotic kinematics, and dynamics. Furthermore, the proposed controller does not require modification of the torque controller, making it applicable for the industrial/commercial robots with closed structure. To address the camera’s capacity for measuring image-space positions and ensure the feature point remains permanently visible, an image-space position-based barrier function (IPBBF) is developed to constrain feature points within the camera’s view directly, rather than utilizing the traditional tracking error-based BF that indirectly converts the image-space position constraints into error constraints, thus eliminating the laborious offline computation process for the parameters of performance functions. Theoretical analysis and experimental results on a robot manipulator are conducted to exhibit the superior performance of the designed method.