dc.contributor.advisor |
Postlethwaite, C |
en |
dc.contributor.advisor |
Egbert, M |
en |
dc.contributor.author |
Jeong, Jiyun |
en |
dc.date.accessioned |
2018-06-18T22:18:00Z |
en |
dc.date.issued |
2018 |
en |
dc.identifier.uri |
http://hdl.handle.net/2292/37303 |
en |
dc.description |
Full text is available to authenticated members of The University of Auckland only. |
en |
dc.description.abstract |
Evolutionary Robotics is a field of science where Darwin’s Evolution Theory is applied to design an artificial system, which has been widely used to study cognitive behaviours. The idea of using a system that couples an artificial agent’s nervous system, body and the environment was introduced in the 1990s, which was followed by the use of a dynamical system to model the nervous system. Continuous-Time Recurrent Neural Networks (CTRNNs) were a common model for the artificial neuron systems for the past two decades, and a variety of behaviours including sensorimotor and learning behaviour were studied in depth using CTRNNs. It is of interest to investigate whether another dynamical system would serve as an evolvable nervous system of an artificial agent. In particular, we are interested in using a heteroclinic network as the nervous system because it is easy to interpret and it can be modelled so that it produces outputs responding to an external input. In this thesis, we use a heteroclinic network as the nervous system of an artificial agent, and couple it to the body of the agent and the environment. We evolve the network using a genetic algorithm, and see if the agent can solve sensorimotor tasks in one dimension and two dimensions. In the one-dimensional problem, the task is to distinguish between a single stimulus and a cluster of stimuli. In the two-dimensional problem, the task is to identify a fixed single stimulus. In both cases, the evolutions were successful. The behaviours of the best-performing agents are analysed to understand how the interactions between the network, the body and the environment make the agent achieve optimal solutions. We also suggest some possible future work in the concluding chapter. |
en |
dc.publisher |
ResearchSpace@Auckland |
en |
dc.relation.ispartof |
Masters Thesis - University of Auckland |
en |
dc.relation.isreferencedby |
UoA99265085013502091 |
en |
dc.rights |
Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated. Previously published items are made available in accordance with the copyright policy of the publisher. |
en |
dc.rights |
Restricted Item. Available to authenticated members of The University of Auckland. |
en |
dc.rights.uri |
https://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htm |
en |
dc.rights.uri |
http://creativecommons.org/licenses/by-nc-sa/3.0/nz/ |
en |
dc.title |
Evolving a Heteroclinic Network in Sensorimotor Tasks |
en |
dc.type |
Thesis |
en |
thesis.degree.discipline |
Applied Mathematics |
en |
thesis.degree.grantor |
The University of Auckland |
en |
thesis.degree.level |
Masters |
en |
dc.rights.holder |
Copyright: The author |
en |
pubs.elements-id |
745041 |
en |
pubs.org-id |
Science |
en |
pubs.org-id |
Mathematics |
en |
pubs.record-created-at-source-date |
2018-06-19 |
en |
dc.identifier.wikidata |
Q112936807 |
|