Wiki Article on RIT Networks

[Sotha]

Due to increased interest in the RIT networks, I've written an extensive wiki article on the subject. It should summarize quite well what I learned when working with the Phrase Book mission.

 

http://wiki.thedarkm...le=RIT_Networks

 

I hope you find it interesting. Enjoy!

[demagogue]

Cool! Anything enhancing more realistic AI behavior is always a plus.

 

I'm writing a wiki entry but finding it hard to boil it down since there's so much I want to say. I like how this one is pretty tight and informative. It's a good model.

[Bikerdude]

Sotha, could you know up a basic map with a simple RIT setup..? I would like to use RIT in my Elixir map to give the impression of more Ai than there actually is.

[Sotha]

http://www.sendspace.com/file/l7mq4e

[Bikerdude]

sweet ta.

[Springheel]

Very nice. I'll definitely be relying on this for my next missions.

[grayman]

One thing to keep an eye out for with RIT when there's a single network that you inject several AI into: one or more AI can end up trying to get to the same path_node at the same time. With lots of places to go, this shouldn't happen too often, but when it does happen, it looks strange. The first one there stands still just fine, but the second one walks round and round the first one, trying to reach the path_node.

 

I had this more often than I liked with two revenants on the same network, so I dup'ed the network and shifted the path_nodes so the revenants weren't trying to reach the same spots.

[i30817]

Could a network edge be marked as "occupied" so it is disqualified from the possibilities (and the percentages adjusted between the rest while it is)?

 

A alternative is to skip the "interresting places" if someone is already there in the same node, but don't prevent going in the same route path (if it looks too mechanistic they are always alone on the routes)

Edited May 17, 2012 by i30817

[grayman]

Could a network edge be marked as "occupied" so it is disqualified from the possibilities (and the percentages adjusted between the rest while it is)?

 

Certainly.

 

But it wouldn't be marked "occupied" until the AI reached it. Until he does, following AI could still choose it, sending one or more toward it.

 

An entrance path_wait could remember its previous choice and not send the next guy there. But that design falls apart when three or more AI are on the same network, unless we let path_waits keep track of who they sent where, but then that falls apart unless we tell a path_wait when a path node becomes "unoccupied", freeing it for someone else.

 

It can get real complex real fast, and probably not worth implementing.

[i30817]

Yeah the first alternative doesn't work. That whole pigeon thing in discrete maths

[PranQster]

I re-did my existing paths in my new map last night, converting them to RIPs. Not only do I like this method much more, but it is far easier to build an initial patrol route. It's also very easy to insert traditional paths into the random mix. I can think of my patrols in a much more modular manner now.

[Professor Paul1290]

I just wanted to say thanks a lot for this. This adds something I kind of always wanted in TDM and in more stealth games in general.

[Sotha]

Fantastic! It is great to see how some experimental work is suddenly turning into solid and lasting technique the mappers embrace. I never dared to expect this fast incorporation of RITs in other FMA's works.

 

Hopefully no unexpected problems emerge...