<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Recent posts to news</title><link>https://sourceforge.net/p/rodney/news/</link><description>Recent posts to news</description><atom:link href="https://sourceforge.net/p/rodney/news/feed.rss" rel="self"/><language>en</language><lastBuildDate>Sun, 29 Jun 2003 20:52:56 -0000</lastBuildDate><atom:link href="https://sourceforge.net/p/rodney/news/feed.rss" rel="self" type="application/rss+xml"/><item><title>New STVM</title><link>https://sourceforge.net/p/rodney/news/2003/06/new-stvm/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;I've added the first version of the new short-term visual memory.  This is based upon earlier VB demos but this time its a C++ DLL.  The STVM is actually a variant of Adaptive Resonance Theory in that the tollerances for exemplars expand or contract dynamically in order to optimally cover the state space.  Instead of just trying to classify the entire image as a gestalt the new memory seeks only to classify salient areas highlighted by motion tracking etc.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Bob Mottram</dc:creator><pubDate>Sun, 29 Jun 2003 20:52:56 -0000</pubDate><guid>https://sourceforge.net7e9106b27c194f281160a012743c16af3a8a960c</guid></item><item><title>Back in a track suit</title><link>https://sourceforge.net/p/rodney/news/2003/06/back-in-a-track-suit/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;It's a long time since I did anything with the sourceforge site, so it was due for an overhaul.  I've dumped most of the earlier stuff (to my horror I couldn't actually delete it, only hide it).&lt;/p&gt;
&lt;p&gt;The new source which I've uploaded is for version 4 of the robot.  Each chunk of code consists of a VC++ 6.0 DLL and accompanying visual basic 6 examples.  I felt that turning most of the critical algorithms into DLLs would simplify the overall software architecture and also make it easier for others to experiment with the system without needing to have a lot of detailed knowledge about how the algorithms work.&lt;/p&gt;
&lt;p&gt;Some of the vision modules are designed to be used in conjunction with one or more webcams.  If you want to try one of these make sure that the drivers for your cameras are WDM compatible and that you enter the correct filter name for the camera within the code.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Bob Mottram</dc:creator><pubDate>Thu, 26 Jun 2003 19:04:58 -0000</pubDate><guid>https://sourceforge.net648495b1057a0e9e61877982426759bfe7302bdf</guid></item><item><title>Stereo vision demo</title><link>https://sourceforge.net/p/rodney/news/2002/05/stereo-vision-demo/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;I've added a stereo vision demo, which shows a very simple and fast method for determining horizontal displacement and distance for neerby objects.  This algorithm seems particularly robust, no matter what type or colour of object is presented.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Bob Mottram</dc:creator><pubDate>Wed, 22 May 2002 21:54:25 -0000</pubDate><guid>https://sourceforge.net8ba4537ef10fb45feda06170ce2755858a22b292</guid></item><item><title>STVM Demo</title><link>https://sourceforge.net/p/rodney/news/2002/05/stvm-demo/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;I've added a standalone demonstration program for the short term visual memory.  You can try it as long as you have a webcam.  The system stores up to 1000 templates called &amp;quot;exemplars&amp;quot; and matches these against the currently observed image.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Bob Mottram</dc:creator><pubDate>Sun, 19 May 2002 16:39:20 -0000</pubDate><guid>https://sourceforge.net0a957236b525a6c3fe8e88536e246cd119a19983</guid></item><item><title>Welcome</title><link>https://sourceforge.net/p/rodney/news/2002/05/welcome/</link><description>&lt;div class="markdown_content"&gt;&lt;p&gt;Rodney is an upper torso humanoid robot designed to investigate the areas of visual perception,  social interaction and large scale semantic networks.&lt;/p&gt;
&lt;p&gt;I've posted the current code for the robot (written in VB6) onto sourceforge in the hope that this will help me go about the software development in a more organised way, and so that other people can see what I'm doing.&lt;/p&gt;
&lt;p&gt;I hope to post more details on the mechanics and construction of the robot which could potentially allow other people to build similar designs.&lt;/p&gt;
&lt;p&gt;I'm also investigating the possibility of porting the code to Linux (Mandrake 8.2 under the C++ Kdevelop IDE) as this may help me to overcome some of the problems which I've been having in running two identical webcams simultaneously.&lt;/p&gt;&lt;/div&gt;</description><dc:creator xmlns:dc="http://purl.org/dc/elements/1.1/">Bob Mottram</dc:creator><pubDate>Sat, 18 May 2002 09:09:05 -0000</pubDate><guid>https://sourceforge.net0ac673b959299b3ac2b334c96a140d7ea2c98b4d</guid></item></channel></rss>