Penggunaan Deep Learning untuk Perhitungan Inverse Kinematics Robot Manipulator 6-DOF (Studi Kasus - Green House Melon ATP IPB CIKARAWANG)

Abstract

In a 6-degree-of-freedom (6-DOF) manipulator robot, an inverse kinematics equation is adopted to determine the joint angle values of the robots based on the desired position and orientation of its end-effector. However, the inverse kinematics solution requires complex computing and only performs well in ideal working environments due to the many 'ideal' assumptions used. In this study, an Artificial Neural Network (ANN) and long short-term memory (LSTM) were used to solve the inverse kinematics problem of a 6-DOF manipulator robot. The robot is designed to meet the requirements of a melon-pruning robot that will be used in the melon greenhouse at Agribusiness and Technology Park (ATP), IPB University. The data for the neural network training were generated from the designed robot using Denavit Hartenberg (DH), and the generated data were used to train the Model with a training-test data ratio of 80:20. The network architecture of the ANN and LSTM was set to 64-128-64, 64-256-64, 64-512-64. The study results show that LSTM performs better, with a loss value of 0.0330, smaller than ANN's (0.0386). This clearly shows that either LSTM or ANN can be used to calculate inverse kinematics for robotic arm manipulators.