Model-Based Control of a 6 DOF Robotic Arm with Visual Recurrent Bilayer ANN

Abstract

This work presents a control model for a 6 DOF robot that includes trajectory generation and tracking. The system recognizes harness assembly ties using an RGB camera that uses a Hopfield bilayer artificial neural network to process data from the assembly plates for assembly zone recognition. Likewise, the kinematic design of the robotic structure to manipulate and control the system is deduced. This document performs a mathematical analysis of the kinematic model that describes the structure, joints, and links, which allows obtaining the xyz coordinates of the end effector. Two control methods are deduced and tested. The Tau-jerk method is integrated with the Newton-Rapson method to obtain the inverse kinematics. Likewise, an adaptive, time-varying algebraic method is used, the control of which does not require square matrices, as well as the integration of an inverse kinematics method.