Meeting with Rem

So i had another meeting with Rem the other day and i must say i found it really cool to sit there and just throw ideas back and forth. Granted my current agent factory knowledge is not up to expert levels but my understanding of how AOPL works has come along to where i can now understand how it works.

We discussed several aspects of the project and i now have a clearer picture of what is needed regarding the later parts of the project but for now i am focusing on simple baby steps.

In the coming weeks i am hoping to do some diagrams to help show what way the project is going to be structured so fingers crossed.

A couple of notes from the meeting.
1) I need to contact Dave Lillis and get set up as a poster to the agent factory wiki
2) I need to get Netbeans and install agent factory and get that working
3) I need to begin to get an agent made that can talk to the SoccerServer.
4) I need to read the newer soccer server manual
5) I need to register for the RoboSoccer mailing list and fire off a few mails to find out all the new improvements to SoccerServer as my current manual is dated 2003
6) I need to read the book chapter Rem gave me
7) I need to read the SoSAA paper
8) I need to set up SVN on Agent Factorys sourceforge

A couple of other important things that came up that i want to note for future reference. The time criticalness of the SoccerServer clients may cause problems later on as agent factory is single thread thus if i have 12 agents and a 400ms cycles each agent only gets roughly 33ms of CPU time. Serveral solutions to this were bounced around during the meeting. I suggested maybe turning off agents when they aren't needed eg the goalie isn't needed if we are attacking so savings like that may help. Also using pure java classes/methods to refine/improve performance in critical parts of the agents programming. Rem even said he would consider making refinements to agent factory so that it was more focused on RoboSoccer/time critical agent based systems. He said he could go back to an earlier version of agent factory that was multi-threaded although he said they found this to be slower than single thread but for my needs it may actually prove faster. Again this is all just being jotted down for later reference.

We also discussed SoSAA which is a method that allowes us to merge CBSE and AOSE ideas. More on this after read the paper rem gave me. Basically CBSE is good at low-level stuff and AOSE is good at high level stuff and S0SAA is a way or merging the aspects of both approaches eg Componets and Agents. But like i said more on this later.

Also Rem was very cool for suggesting that we may release some of my stuff on the agent factory wiki when it is a viable/useful tool.

So that all for now i have a tonne of college work due between now and the end of term I'll cross off what i get done form the list when i get it done and hopefully i can start to make a bit of progress between now and the end of January.

Agent Factory ON HOLD

Yeah so i've been really busy and will be for the next couple of weeks with course work. If i get some free time between now and christmas i'll blog about it but don't expect anything new till the new year :(