Adaptive robotic arm control using artificial neural network

• Main Authors: Ligutan, Dino Dominic F., Abad, Alexander C., Dadios, Elmer Jose P.
• Format: text
• Published: Animo Repository 2019
• Subjects: Robots—Control systems; Robot hands; Neural networks (Computer science); Controls and Control Theory; Electrical and Electronics
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/1906
• Institution: De La Salle University

Robots have been used for exploration where human lives may be endangered. One such robot is the bomb disposal robot. Part of bomb disposal robot is the robotic arm needed to inspect suspected items. This study demonstrates the use of artificial neural networks (ANN) in control of a 4-DOF robotic arm with a 2-DOF gripper end-effector. The robotic arm joint space is sampled at regular intervals and the coordinates of the end-effector are acquired for each of these points. These joints and coordinates are preprocessed to form the approximate Jacobian of the robotic arm and is fed to an artificial neural network. The controller design is based upon the Jacobian transpose method. Results indicate the ability of the designed controller to compensate for loading effects with an absolute error of no more than 5 millimeters from the desired target. © 2018 IEEE.