THE FORWARD KINEMATICS PROBLEM OF CABLE-DRIVEN PARALLEL ROBOTS

Abstract

Kinematic model, including inverse kinematics (IK) and forward kinematics (FK), is the first problem that needs to be solved when researching cable-driven parallel robots (CDPR). In contrast with inverse kinematics, where the equations are decoupled and can be solved directly, the forward kinematics problem is more involved. Generally, the forward kinematics problem of CDPR are not analytically solvable. However, for a general CDPR with six degree of freedom (DoF), there is no analytical solution at hand. Therefore, numerical method has to be considered to find the solution, which is a drawback in terms of time consumption, especially in real-time computing. In this study, to solve the problems related to the forward kinematics problem, we propose a neural network to estimate the position and orientation of end-effector (EE) based on the shallow learning method. The input training data is the length of the cables and the output data is the position and orientation of EE obtained from the inverse kinematic model.