Well, thnx to RPGeezus as his coverage of POD bot is pretty insightful. While this may begin to get al little off topic, I think it''s important to cover as POD bot is a great example of fps a.i.
If you plan on playing around with POD bot, I would suggest using the radio commands, as some of those are very neat and even helpful, things like this should be used more often with a.i. as communication usually leads to a better game.
Here''s some simple answers to your comments on POD bot, RPGeezus.
1) one way to aviod crowding of bots is to have a min. distance allowed between bots, as this has advantages and disadvantages. This would obviously need to be tweeked.
as for the bots'' jump and shoot behavior, this can be fixed by setting the bots to do a different action, randomly for a more believable effect. My DoD bots, since the maps offer cover everywhere, will crouch if they spot an enemy and pop out to take some shots. This looks VERY realistic, but eventually it just looks dumb because I don''t have ''em set to do something else besides crouch.
2) Avoid waypoints are quite pointless,if you don''t want a bot to go somewhere in the level, then don''t place waypoints there for them to move in that position.
3) as for locational memory of enemy positions, I haven''t gotten this in yet and I haven''t seen this used really, but POD bot has a function where the bot will hide and then pop out about 4 sec. later. But this isn''t what u wanted i''m sure, but I''ll get there.
And waypoint priority does exist, just not in the manner you mention. This could be done though, and would be a nifty little feature to add, i''ll have to try that out.
4)Firing lanes are a much more complicated thing to create, and making them "aware" of what enemies are using as far as guns are concerned, i don''t think this will be appearing any time soon, making the bot recognize the weapon and using appropriate actions for that situation would be a very challenging thing to do. But worth it I''m sure. However, if bots are fighting by another, and they realize that their buddies are dying by the enemy, they could be given a feature that tells them to select a new path.
5)For POD, if the radio command "get in position" was modified to tell the team to pick set camping spots, then a basic defence could be made, rather than the random placement of camping spots that''s currently being used.
currently, what i''ve noticed with POD is that they do infact stay in position for a time and they roam about, but they select a new firing position. Perhaps if this timer was changed to a longer period, the bots wouldn''t move as much when defending.
a great example of all of this is de_cbble. Be a counter-T and use the command "get in position" to start, and then "go, go, go!" when the bomb is planted. u''ll see the T''s in defending positions, and minor movement in an attempt to get a bead on you.
The POD bkot source code is available, so if you want, you can make those minor changes, and surely get some good enjoyment out of them.
I''ll be working on those DoD bots of mine and try to get them to work a little more like POD. When I''m done with the bots, I''ll be sure to let you all know about it before I submit them to planet hal-life and so forth.
"you can''''t understand some thing you can''''t create."
Will A.I. for FPS always stay simple?
When You run in a room with only one exit , at least on all the games I play, all the boots Just run in and shot. This looks realy stupid. The bots should wait outside the door to come out or should at least throw in a greanade while only exposizing and arm or so. Or better yet light something on fire like some trash or the celling.And breathing it hurts you.
In fact any time the enemy knows your mission. Sometimes you are trying to infueltrate the enemy base and they know this somewhere on the level they should set an ambush make the door look ungaurded but have gauds in anouther room not in the direct path run out and encirle you.
At the 2002 GDC, I attended an all-day tutorial given by John Laird - father of the SOAR programming language. He was showing us some examples of how his Quake bots worked.
In the first example, he was talking about how when his bot maps out a room (or collection of them) it will determine a camping spot. For example, imagine a square or rectangular room with entrances in the middle of the east wall and the middle of south wall. The bot would tend to move to the southeast corner and then face toward the center of the room. The reason is, that is the only corner that can be seen from none of the doorways.
In another example, it showed some advanced planning. Imagine a room where there is a door from the west and a door to the north. The door to the north leads down a hall to a small room with a known power-up. When chasing an enemy into the room from the west, and the enemy is no longer in sight, the bot would realize that the enemy has gone out the north door. Rather than chase it down that hall to the power-up, the bot would camp in the southwest corner facing toward the north door. It would have calculated approximately how long it would take for the enemy to go up the hall, grab the power-up, and return back down the hall to the main room. The bot would be polite enough to fire right about the time when the enemy should be walking back into the room!
THAT is the sort of AI that would make FPS games more involving.
Dave Mark - President and Lead Designer
Intrinsic Algorithm - "Reducing the world to mathematical equations!"
In the first example, he was talking about how when his bot maps out a room (or collection of them) it will determine a camping spot. For example, imagine a square or rectangular room with entrances in the middle of the east wall and the middle of south wall. The bot would tend to move to the southeast corner and then face toward the center of the room. The reason is, that is the only corner that can be seen from none of the doorways.
In another example, it showed some advanced planning. Imagine a room where there is a door from the west and a door to the north. The door to the north leads down a hall to a small room with a known power-up. When chasing an enemy into the room from the west, and the enemy is no longer in sight, the bot would realize that the enemy has gone out the north door. Rather than chase it down that hall to the power-up, the bot would camp in the southwest corner facing toward the north door. It would have calculated approximately how long it would take for the enemy to go up the hall, grab the power-up, and return back down the hall to the main room. The bot would be polite enough to fire right about the time when the enemy should be walking back into the room!
THAT is the sort of AI that would make FPS games more involving.
Dave Mark - President and Lead Designer
Intrinsic Algorithm - "Reducing the world to mathematical equations!"
Dave Mark - President and Lead Designer of Intrinsic Algorithm LLC
Professional consultant on game AI, mathematical modeling, simulation modeling
Co-founder and 10 year advisor of the GDC AI Summit
Author of the book, Behavioral Mathematics for Game AI
Blogs I write:
IA News - What's happening at IA | IA on AI - AI news and notes | Post-Play'em - Observations on AI of games I play
"Reducing the world to mathematical equations!"
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
Recommended Tutorials
Advertisement