Neural network in solving inverse kinematic near singularity of a two-joint arm manipulator

Name: Akapot Tantrapiwat

Organization : King Mongkut Institute of Technology Ladkrabang. Faculty of Engineering

Email : akapot.ta@kmitl.ac.th

keyword: Singularity Points

ThaSH: Manipulators (Mechanism) -- Automatic control

; Inverse Kinematics

ThaSH: Neural networks (Computer science)

; Two-joint Arm Manipulator

ThaSH: Kinematics

Abstract: A simulation of two-joint arm manipulator made in Python programming was created and studied in order to search for its inverse kinematic solutions using artificial neural networks. This manipulator configuration is similar to the SCARA robot arm without the Z-axis and the End-Effector orientation. The research was aimed to compare different sizes of neural networks and training epochs which affect the number of trials in searching for a solution near the singularity points which is more difficult in comparison to other locations. The evaluation was made between the six neuron and ten neuron systems with two hidden layers for both platforms. With different training epochs, the results have shown that both neural networks can perform equally on the locations away from the singularity. With the increment of training numbers, the ten neuron system can perform much better when moving between points near the singularity.

King Mongkut's University of Technology North Bangkok. Central Library

Address: BANGKOK

Email: library@kmutnb.ac.th

Created: 2018

Modified: 2025-06-26

Issued: 2025-06-26

บทความ/Article

application/pdf

BibliograpyCitation : In Thammasat University Faculty of Engineering. The 9th TSME International Conference on Mechanical Engineering 2018 (TSME-ICoME9) (pp.562-568) Pathum Thani : Thammasat University

eng

©copyrights King Mongkut's University of Technology North Bangkok

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	TSME-ICoME9 2018pp.562-568.pdf	1.23 MB

ใช้เวลา

0.118772 วินาที