Joint Constraint Modelling Using Evolved Topology Generalized Multi-Layer Perceptron(GMLP)

Jenkins, Glenn and Angel, Paul (2008) Joint Constraint Modelling Using Evolved Topology Generalized Multi-Layer Perceptron(GMLP). International Journal of Simulation Systems, Science and Technology. ISSN 1473-8031

[img]
Preview
Text
Jenkins2008.pdf
Available under License Creative Commons Attribution.

Download (468kB) | Preview

Abstract

The accurate simulation of anatomical joint models is important for both medical diagnosis and realistic animation applications.  Quaternion algebra has been increasingly applied to model rotations providing a compact representation while avoiding singularities.  This paper describes the application of artificial neural networks topologically evolved using genetic algorithms to model joint constraints directly in quaternion space.  These networks are trained (using resilient back propagation) to model discontinuous vector fields that act as corrective functions ensuring invalid joint configurations are accurately corrected.  The results show that complex quaternion-based joint constraints can be learned without resorting to reduced coordinate models or iterative techniques used in other quaternion based joint constraint approaches.

Item Type: Article
Additional Information: Citation: Jenkins G., Angel P., Joint Constraint Modelling Using Evolved Topology Generalized Multi-Layer Perceptron (GMLP), International Journal of Simulation Systems, Science and Technology, Volume 9, No 5. pp. 15-26 December 2008, ISSN: 1473-8031.
Uncontrolled Keywords: Anatomical joint constraint, unit quaternion, piece-wise linear function approximation, discontinuous vector field, evolved topology neural network, NetJEN
Subjects: R Medicine > R Medicine (General)
Divisions: Faculties > Faculty of Architecture, Computing and Engineering > School of Applied Computing
Depositing User: John Dalling
Date Deposited: 25 Apr 2012 15:59
Last Modified: 28 Jan 2016 16:23
URI: http://repository.uwtsd.ac.uk/id/eprint/258

Actions (login required)

View Item View Item