Autonomous Data-Driven Manipulation of Unknown Anisotropic Deformable Tissues Using Unmodelled Continuum Manipulators

Abstract

We present an autonomous manipulation approach for tissues with anisotropic deformation behavior using a continuum manipulator. The key feature of our vision-based study is an online learning and estimation method, which makes its implementation independent of any prior knowledge about the deformation behavior of the tissue and continuum manipulator as well as calibration of the vision system with respect to the robot. This important feature addresses the difficulty of using model-based control approaches in deformation control of a continuum manipulator manipulating an unknown deformable tissue. We evaluated the performance and robustness of our method in three different experiments using the da Vinci Research Kit coupled with a 5 mm instrument that has a 4-degree-of-freedom snake-like wrist. These experiments simulated situations that occur in various surgical schemes and verified the adaptability, learning capability, and accuracy of the proposed method.