This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License
|
||||||||
|
Paper Details
Paper Title
Application of RBFN for forward kinematics solution of a parallel manipulator
Authors
  Dr. Sudipto Chaki
Abstract
In the present work different architectures of Radial Basis Function Networks (RBFN) with supervised learning are trained and tested for solving forward kinematics (FK) of a 3-3 UPU manipulator which can predict end effector location uniquely for real time implementation of the manipulator. Performances of the architectures during training and simulation are also analysed in detail and its superiority over traditional backpropagation is also established. An attempt is also taken to find empirical equations to predict the error during training and simulation with variation in spread and fixed hidden layer neurons in RBFN for the present problem. Performance of the network for tracking a trajectory in a plane is also shown. Ultimately it is found that, the best architecture among tested can predict any arbitrary location in the workspace of the manipulator with maximum errors and average errors of 0.018339 cm and 0.001860 cm during simulation. As errors during simulation are confined only within a very small range it can be easily applied in real time applications.
Keywords- Forward kinematics (FK), 3-3 UPU Manipulator, Radial Basis Function Networks (RBFN)
Publication Details
Unique Identification Number - IJEDR1902091Page Number(s) - 478-483Pubished in - Volume 7 | Issue 2 | June 2019DOI (Digital Object Identifier) -    Publisher - IJEDR (ISSN - 2321-9939)
Cite this Article
  Dr. Sudipto Chaki,   "Application of RBFN for forward kinematics solution of a parallel manipulator", International Journal of Engineering Development and Research (IJEDR), ISSN:2321-9939, Volume.7, Issue 2, pp.478-483, June 2019, Available at :http://www.ijedr.org/papers/IJEDR1902091.pdf
Article Preview
|
|
||||||
|