Shape Sensing of Flexible Manipulators With Visual Occlusion Based on Bezier Curve

Flexible manipulators are promising in minimally invasive surgery, because they can work well in complex and confined environment. The size of the manipulator is usually small, which makes it very difficult to install sensors to measure the distal tip position and manipulator shape. However, the shape information is crucial in controlling the instruments. In this paper, the shape of the manipulator is estimated by endoscopic view based on Bezier curve and the Levenberg-Marquardt algorithm. This method works even the view is partially occluded. Compared with existing none-vision based shape sensing method, no additional sensor is needed in the proposed method. The method is evaluated by both simulations and experiments. Meanwhile, the influence of the occlusion on the sensing accuracy is analyzed. The experimental results show that the shape information of the flexible manipulator both with and without payload can be well estimated by the proposed method.