Model-Free and Uncalibrated Eye-in-Hand Visual Servoing for Magnetic Suspension and Lesion Tracking of Capsule Robots

Abstract

This article proposes a model-free and uncalibrated eye-in-hand (EiH) visual servoing approach for controlling magnetically-actuated capsule endoscopy in a suspended state with an external permanent magnet fixed at the end-effector of a serial-link robot manipulator (in GI tract diagnosis). We implement active motion control of the capsule endoscopy by visual feedback through feature information in the image plane to correct the position of the robot manipulator’s end-effector. This method decouples motion control in the z-axis direction from motion control in the x- and y-axis directions, ensuring a rapid response in suspension control. Image feature deviations have been mapped to robot actuation variables with image Jacobian estimated by unscented Kalman filter to control the motion in the x- and y- direction. In the meantime, a PD controller has been utilized to control the motion in the z-direction. Experiments conducted in liquid environments confirm the effectiveness of the algorithms for suspension control and lesion tracking in capsule robot.