Eye-In-Hand Uncalibrated Visual Servoing of Concentric Tube Robot

Abstract

A control strategy of uncalibrated visual servoing is proposed for a three-degree-of-freedom continuous concentric tube robot (CTR) in the paper. CTRs are widely researched in recent years for its advantages of flexibility, minimally invasive and portable medical device surgery. Compared with calibrated visual servoing control, uncalibrated visual servoing does not require CTR's accurate robot model and camera calibration parameters, while the difficulty of uncalibrated visual servoing is how to accurately estimate the Jacobian matrix of the image online. Therefore, the paper designs a control scheme of uncalibrated visual servoing for a three degree-of-freedom CTR. Kalman Filtering(KF) and Unscented Kalman Filtering(UKF) algorithms are used to estimate the Jacobian matrix of the CTR system online. A millimeter level endoscope is installed at the end of the CTR's operating arm for image acquisition, which is a an eye-in-hand visual system. The best UKF parameters for the CTR are obtained based on comparative experiments. Experiments show that the strategy of uncalibrated visual servoing based on designed UKF algorithm has a good performance in locating and tracking the target point for the CTR.