Remarks on Learning Inverse Kinematics of a Robot Manipulator Using a Quaternion Neural Network

Abstract

In this study, the application of a quaternion neural network (QNN) for solving the inverse kinematics problem of a robot manipulator is investigated. To obtain the solution of the inverse kinematics problem, the QNN earns the mapping between the work space and the joint spaces of the robot manipulator through training on a dataset predetermined by the forward kinematics of the robot manipulator. A seven degree-of-freedom robot manipulator is employed in the computational experiments and the simulation results demonstrate the feasibility of using the QNN for this task.