Maiden Project
Hi. Sorry about the long delay.
I started work on the Maiden project, a sort of lovechild of Erc and its example code.
I found out early on that Erc was very good for AI coding. In fact, I was able to do complex AI problems not only in much fewer lines, but also without at much necessary previous planning. The Erc structure and syntax allows one to easily visualize a solution to an AI problem and provides an excellent environment to describe and execute a solution to such a problem. With the advent of revision 1.1, I found myself with a few people who were interested in it (though mostly from my shameless promotion on IRC and USENET), and I started writing example code. The example code, while short, illustrated how easily Erc could be applied to AI problems. The logical parser was written in 11 lines, and was able to implement the chain rule to deduce relationships and provide all the possible values for a statement to be true given multiple possibilities. I wrote a natural language parser core is about 15 lines, which allowed for translation of a given natural language (assuming the correct syntaxpack is loaded) to be translated into equivalent Erc code, which then could hypothetically be executed and the result translated back into the natural language.
With these examples in hand, I decided to take the next step. I decided to make a robotics/AI core module, extensible in Erc, for use either as a Linux kernel module or an embedded OS. This is the Maiden project.
I began work yesterday, writing headers, documentation, and makefiles, and continued today with the beginning of the module piece. I have registered a project on SourceForge for the Maiden project, and hope that it will be approved. I made a logo yesterday as well. Jon will likely help me, albeit with some resistance, and Two9a has agreed to try out helping.
I started work on the Maiden project, a sort of lovechild of Erc and its example code.
I found out early on that Erc was very good for AI coding. In fact, I was able to do complex AI problems not only in much fewer lines, but also without at much necessary previous planning. The Erc structure and syntax allows one to easily visualize a solution to an AI problem and provides an excellent environment to describe and execute a solution to such a problem. With the advent of revision 1.1, I found myself with a few people who were interested in it (though mostly from my shameless promotion on IRC and USENET), and I started writing example code. The example code, while short, illustrated how easily Erc could be applied to AI problems. The logical parser was written in 11 lines, and was able to implement the chain rule to deduce relationships and provide all the possible values for a statement to be true given multiple possibilities. I wrote a natural language parser core is about 15 lines, which allowed for translation of a given natural language (assuming the correct syntaxpack is loaded) to be translated into equivalent Erc code, which then could hypothetically be executed and the result translated back into the natural language.
With these examples in hand, I decided to take the next step. I decided to make a robotics/AI core module, extensible in Erc, for use either as a Linux kernel module or an embedded OS. This is the Maiden project.
I began work yesterday, writing headers, documentation, and makefiles, and continued today with the beginning of the module piece. I have registered a project on SourceForge for the Maiden project, and hope that it will be approved. I made a logo yesterday as well. Jon will likely help me, albeit with some resistance, and Two9a has agreed to try out helping.
0 Comments:
Post a Comment
<< Home