ChrisTingley
Articles
An Introduction to Ruby on Rails
Presented at the 1st Reading Geek night, 11th Novemeber 2009
"He showed us some 'speed coding' with a live demo of Ruby on Rails. In the space of 15 mins he wrote a web app from scratch (including writing some test cases) and gracefully recovered from an (extremely rare) mac related crash! Chris is clearly a Ruby ninja and opened a few eyes to what you can do with an MVC framework like Rails." [reference]
Chris Tingley CV
2009, Available as downloadable Document (70kB). Downloaded 1600 times
My CV
Working towards an emotion-based agent architecture
2006, Available as downloadable Paper (200kB). Downloaded 164 times
It is known that emotions provide useful functions within animals and humans. It has been proposed that agent systems can also benefit from biologically inspired emotions, but to do this an adequate emotional-based architecture must be developed. This project has developed such an architecture using a test platform consisting of three physical robots and a number of virtual agents. From observing the emergent behaviors of the agents equipped with the developed emotion-based architecture, several interesting and beneficial traits were identified. 1) A reduced complexity in behavioral programming and action selection due to the abstraction of inputs in which turn leads to simpler creation of emotional responses. 2) Reduced bandwidth between the agents and controllers and 3) more realistic behaviors.
Developing an emotion-based agent architecture and Development Platform
2006, Available as downloadable Technical Report (1MB). Downloaded 112 times
Abstract: This report introduces the importance of emotions and their uses in control mechanisms for autonomous agents, but in order to develop such architectures, a suitable development platform is needed. This project not only tackles the problem of laying the foundations for a comprehensive emotion-based agent architecture, but also addresses the need to have an extensible platform to develop and test both physical and virtual implementations of the developed agent architectures. The project has successfully culminated in a reliable mixed reality environment where physical and virtual agents can coexist and interact, whilst providing a solid base for a number of other projects using the same installation. Alongside this, an architecture has been developed providing a number of benefits which are outlined in the results in section 3 including (although not limited to) simpler behavioral programming and reduced bandwidth requirements between agents and controllers.
Exploring the needs of Emotions in a Multi-Agent System
2006, Available as downloadable Research Paper (1MB). Downloaded 159 times
Human emotions have fascinated scientists for many years because of the diversity of findings between differing fields of study. Even though there is no clearly defined answer as to what emotions are and why we have them; we can still apply our limited knowledge of emotions to other areas of science. Distributed artificial intelligence; specifically Multi-Agent Systems (MASs), is a relatively new area of science that has only come about within the last 10 years. As a result of its' infancy, many MASs suffer from a number of common fundamental problems. Planning, scheduling and goal management are all inter-related problematic areas of MASs. These are caused by large search domains, multiple dimensionality, non-linearity, conflicting interests and NP-complete problems, all of which are seen to be echoed, but more importantly solved in human systems. This paper will show how human systems overcome these problems using simple emotional mechanisms. A hypothesis will be made that many challenges within MASs can be solved by implementing mechanisms which mimic the purpose of human emotions.
Developing an emotion-based agent architecture and development platform
2006, Available as downloadable Technical Report (49MB). Downloaded 159 times
This report introduces the importance of emotions and their uses in control
mechanisms for autonomous agents, but in order to develop such architectures, a suitable
development platform is needed. This project not only tackles the problem of laying the
foundations for a comprehensive emotion-based agent architecture, but also addresses the
need to have an extensible platform to develop and test both physical and virtual imple-
mentations of the developed agent architectures. The project has successfully culminated
in a reliable mixed reality environment where physical and virtual agents can coexist and
interact, whilst providing a solid base for a number of other projects using the same instal-
lation. Alongside this, an architecture has been developed providing a number of bene?ts
which are outlined in the results in section 3 including (although not limited to) simpler
behavioral programming and reduced bandwidth requirements between agents and con-
trollers.
Cybernetics Tech Report Latex Files
2006, Available as downloadable Files (1MB). Downloaded 150 times
UPDATED! Version 4. Fixed a few more bugs!
A must have for anyone doing a cyb technical report in Latex. Includes an example .tex and the all important .sty. Thanks to Robert Oates and John Wesley for providing version 1.
Cybernetics SCARP Latex Files
2006, Available as downloadable Files (1MB). Downloaded 96 times
Latex style files for SCARP paper submission. Note: there still remain some issues with this style file, please see the example for workarounds.
Evolutionary Computation Coursework
2005, Available as downloadable Files (222kB). Downloaded 146 times
This zip file contains the C#.net projects and source code suitable to implement a generic evoluationary algorithm (specifically a Genetic Algorithm) to solve either: Part A - single peak function, Part B: A multi-peak function, Part C: the Iterated Prisoners Dilemma. See the accompanying PDF for more details.
An Introduction to: Evolutionary Computation
2005, Available as downloadable Technical Report (1MB). Downloaded 195 times
Evolutionary computation is a broad name for several incarnations of evolutionary algorithms (EAs). In general, EAs maintain a population of individuals which evolve based on a number of stochastic and algorithmic techniques applying to the selection, recombination and mutation of each individual within the (often randomly) created population. Several different types of EA will be explored, their similarities and differences illustrated, before implementations of two EAs are described and the influence of varying techniques and parameters on their results are discussed.
Highly Integrated Navigation Algorithm for Alfresco Mobile Robots
2005, Available as downloadable Paper (172kB). Downloaded 291 times
Abstract: A team of six were formed to compete in the 2005 MEng. Robot Challenge and as such build a robot to autonomously navigate between a series of GPS (WGS84) coordinates. This paper discuses the author’s work in designing and implementing a highly integrated navigation algorithm to marshal data between numerous types of sensors and actuators. The intermediate results of the project will be explored and a successful start to the project will be ascertained.
Mobile Robot Project
2004, Available as downloadable Technical Report (1MB). Downloaded 322 times
Abstract: Mobile robots have many uses in today’s modern society and are creeping more and more into the public eye. This project is aimed to give an understanding some of the basic principles in designing and building such robots. The mobile robot produced for this project had one simple goal in life… to follow an object at a fixed distance.
Content-based language analysis MLP
2004, Available as downloadable Files (2.5MB). Downloaded 178 times
A zip file containing source code (and library MLP code) and XCode project for a MLP Neural Network designed to learn and classify text files in different languages. See the accompanying report for more details, or the readme file for useage information.
Using Multilayered Perceptron Artificial Neural Networks in Content-based Language Classification
2004, Available as downloadable Technical Report (3.9MB). Downloaded 156 times
Abstract: Multilayered perceptrons (MLP’s) are a form of artificial neural network (ANN), which are particularly good at classification problems. Content-based language classification (CBLC) is one such example of a classification problem. This paper will look at the design methodology behind creating an MLP network in C++ and show the testing of the network on a so called ‘hard problem’. The paper will finally discuss the implementation of CBLC and show how a trained network can successfully classify up to six European languages; with an accuracy of up to 99.8% on unseen data.