Comments on: Dynamic NPC Actions http://charlessolar.com/post/16 Wed, 02 Jun 2010 19:05:06 +0000 hourly 1 By: Charles Solar http://charlessolar.com/post/16/comment-page-1#comment-16 Charles Solar Sun, 03 Jan 2010 00:39:21 +0000 http://www.mrlwork.com/?p=16#comment-16 Hi mark. I recently wrapped up another prototype project based on my initial theory. You are right though, it has been a long time. Trouble is I run into these road blocks that completely break the design and I have to start back at square one. Take my initial design for example, while I was writing a demo app I realized that there was no real way for the controllers to be able to pick out certain data from the actor without breaking encapsulation. Because I wanted to extend this design into intelligent actions and even automatic dialog <- this would be huge, I had to jump back and rethink some things. To prototype my theories I have tried writing quite a few demo apps where the plain of controllers are letters and its the network's job to parse and write back words and phrases. The idea is if I can make the program talk I can make it walk, sort of speak. Anyway my latest prototype will parse words and build the network, but I ran into yet another road block when it came time to judge actions it can take. The network builds, but when the program is confronted with a choice between executing one tree of controllers or a separate similar tree, it cannot decide on its own which one is better. Its been a very interesting 2.5 years, but I am afraid I have nothing physical to present yet. I have an idea for a judge system but I have yet to write it into the project. Thanks for commenting Hi mark.
I recently wrapped up another prototype project based on my initial theory. You are right though, it has been a long time.
Trouble is I run into these road blocks that completely break the design and I have to start back at square one. Take my initial design for example, while I was writing a demo app I realized that there was no real way for the controllers to be able to pick out certain data from the actor without breaking encapsulation.
Because I wanted to extend this design into intelligent actions and even automatic dialog <- this would be huge, I had to jump back and rethink some things.

To prototype my theories I have tried writing quite a few demo apps where the plain of controllers are letters and its the network’s job to parse and write back words and phrases. The idea is if I can make the program talk I can make it walk, sort of speak.

Anyway my latest prototype will parse words and build the network, but I ran into yet another road block when it came time to judge actions it can take. The network builds, but when the program is confronted with a choice between executing one tree of controllers or a separate similar tree, it cannot decide on its own which one is better.

Its been a very interesting 2.5 years, but I am afraid I have nothing physical to present yet. I have an idea for a judge system but I have yet to write it into the project.

Thanks for commenting

]]> By: Mark http://charlessolar.com/post/16/comment-page-1#comment-15 Mark Sun, 03 Jan 2010 00:16:36 +0000 http://www.mrlwork.com/?p=16#comment-15 So how is this project going? Its been what, 2.5 years? I'm interested :) So how is this project going? Its been what, 2.5 years? I’m interested :)

]]>