thanks to BU
 

Hi!
I've done (better to say just started) a bot for Joint Operations (nothing to do with HL games), and the stuff on this site were very helpful. I've used no code, but the concepts (A*, autowaypointing) i could nicely adopt.
I'm into higher stuff now, like commanding, squads, vehicles, and evasion-logic.

I have an idea on using A* node-costs for making bots evade players. I record when a player makes a kill, and where was the player and the killed enemy standing at that time. The next time when the player steps on a node, i temporarily rise the costs of all the affected nodes, so bots will evade those nodes. The JO game is much different from HL, but have someone done such a juggling with the node-costs? Have it worked?

Re: thanks to BU
 

Nice to hear :)
Some day I will share my deep knowledge of bot coding aswell on the wiki 9_9

For the A* costs...currently the cost for my nodes are only calculated by how much space around the node exist(so bots will choose open paths...). But this doesn't have to be good for tactical games where you have to hide instead :)

Also keep in mind that your way might be considered as 'cheating'. Since the bots know where its dangerous without the need to see/experience it first.
Maybe its better to have a node list for every bot, so each bot can add cost to the nearest node when he dies. That way every bot can learn where its dangerous or not :)

Re: thanks to BU
 

Maybe you could collect a "Using A* 1001 ways" list. I just love this algorithm, its almost like a programming philosophy. :D

I combine 5 factors into the cost, but i'm still having problems on balancing those.

Re: thanks to BU
 

I'm not surprised. Balancing A* is one of the most subtle and most boring tasks an AI programmer has to do. Good luck! :)