Spatial Shape Estimation of a Tendon-Driven Continuum Robotic Arm Using a Vision-Based Algorithm

Abstract

Due to the nonlinearity of the continuum manipulators and the absence of direct relation between the actuator space and task space, shape configuration cannot be achieved by knowing the actuator position. To resolve this problem, an estimation algorithm based on vision processing is proposed in this research. In order to estimate the shape of a two-section tendon-driven continuum robotic arm, each section is considered as a circular arc of multiple curvature in the 3D space. To implement the proposed method, few LEDs are implanted along the length of the manipulator; whereas two webcams are used to capture the photos of LEDs which are exploited to determine their position. These positons are next used for shape estimation using the proposed algorithm. Finally, the efficiency of the proposed algorithm is experimentally evaluated on a calibration setup of the robot and the results are discussed.