Artificial Neural Networks Applied in the Solution of the Inverse Kinematics Problem of a 3D Manipulator Arm

Abstract

This paper approaches the classic problem of robotics' Inverse Kinematics. This problem does not have a trivial solution due its solution complexity, in addition to multiple solutions for the same final position of the robotic actuator. The number of solutions increases with the number of degrees of freedom In this context, the goal of this paper is to present a solution to this problem through the application of an Artificial Neural Network (ANN) in a robotic arm with 4 degrees of freedom in three dimensions. Initial results with a set of seven points using two different ANN configurations were convincing.