Learning Global Inverse Kinematics Solutions for a Continuum Robot

This paper presents a learning based approach for obtaining the inverse kinematics (IK) solution for continuum robots. The proposed model learns a particular global solution for IK problem by supervised learning without any prior knowledge about the system. We have developed an approach that solely relies on the sampling method and a unique IK formulation. The convergence of the solution, practically feasible sample data requirements and adaptability of the model is shown with simulations of a redundant continuum robot.