Perbandingan Solusi Permasalahan Inverse Kinematics pada Robot KUKA KR 5 Arc dengan Metode Neural Network, Recurrent Neural Network, dan Long Short-term Memory

Ramadhani, Rahmat (2023) Perbandingan Solusi Permasalahan Inverse Kinematics pada Robot KUKA KR 5 Arc dengan Metode Neural Network, Recurrent Neural Network, dan Long Short-term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07111940000046-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 September 2025. Download (4MB) | Request a copy

Abstract

Robot manipulator adalah robot yang diprogram untuk menjalankan tugas tertentu dengan cepat, efisien, dan akurat. Forward kinematics dan inverse kinematics merupakan 2 hal yang sering dibahas pada robot manipulator. Permasalahan forward kinematics dianggap tidak terlalu sulit untuk diselesaikan dan tidak terlalu rumit jika dibandingkan dengan inverse kinematics karena penyelesaian permasalahan inverse kinematics memiliki banyak solusi jika dibandingkan forward kinematics. Machine learning sering digunakan dalam menyelesaikan permasalahan inverse kinematics. Pada penelitian ini dilakukan percobaan serta membandingkan antara solusi inverse kinematic yang diperoleh menggunakan metode Neural Network, RNN (Recurrent Neural Network), dan LSTM (Long Short-Term Memory) pada robot KUKA KR 5 Arc yang memiliki 6 derajat kebebasan. Hasil dari penelitian ini menunjukkan bahwa solusi dengan metode Neural Network mampu memberikan solusi yang lebih bai dengan nilai evaluation loss (RMSE) 0,204 derajat jika dibandingkan dengan solusi yang diperoleh menggunakan metode RNN dan LSTM dengan nilai evaluation loss (RMSE) secara berututan 6,096 derajat dan 5,864 derajat. Hal ini juga didukung dari hasil pengujian membuat pola octagon dengan masing-masing model. Pada percobaan dengan neural network diperoleh rata-rata error posisi sebesar 2,217mm dan error orientasi sebesar 0,190 derajat. Pada percobaan dengan RNN diperoleh rata-rata error posisi sebesar 11,999mm dan error orientasi sebesar 0,617 derajat. Pada percobaan dengan LSTM diperoleh rata-rata error posisi sebesar 56,525mm dan error orientasi sebesar 1,419 derajat.
================================================================================================================================
A robot manipulator is a robot that is programmed to perform certain tasks quickly, efficiently, and accurately. Forward kinematics and inverse kinematics are 2 things that are often discussed in robot manipulators. Forward kinematics is considered not too difficult to solve and not too complicated when compared to inverse kinematics because solving inverse kinematics problems has many solutions when compared to forward kinematics. Machine learning is often used to solve inverse kinematics problems. In this study, an experiment was conducted and a comparison was made between the inverse kinematic solutions obtained using the Neural Network, RNN (Recurrent Neural Network), and LSTM (Long Short-Term Memory) methods on the KUKA KR 5 Arc robot which has 6 degrees of freedom. The results of this study indicate that the solution using the Neural Network method is able to provide a better solution with an evaluation loss (RMSE) value of 0.204 degrees when compared to the solutions obtained using the RNN and LSTM methods with an evaluation loss (RMSE) value of 6.096 degrees and 5.864 degrees, respectively. This is also supported by the test results for making octagon patterns using each model. In the experiment with the neural network, the average position error was 2.217mm and the orientation error was 0.190 degrees. In the experiment with the RNN, the average position error was 11.999mm and the orientation error was 0.617 degrees. In the experiment with LSTM, the average position error was 56.525mm and the orientation error was 1.419 degrees.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	inverse kinematics, KUKA KR 5 Arc, long short-term memory, neural network, recurrent neural network
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > TJ Mechanical engineering and machinery > TJ211 Robotics. T Technology > TJ Mechanical engineering and machinery > TJ211.4 Robot motion T Technology > TS Manufactures > TS227 Welding.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Rahmat Ramadhani
Date Deposited:	26 Jul 2023 03:11
Last Modified:	26 Jul 2023 03:11
URI:	http://repository.its.ac.id/id/eprint/99300

Actions (login required)

View Item