GoldenEye: Source Forums
Debriefing => General Goldeneye => Topic started by: John on February 19, 2010, 07:05:59 pm
-
Hi,
I want to know if GES will be able to be play with bots ?
Thx.
See Ya.
-
Not currently, though bots might be implemented in the future.
-
We're working on it. Hopefully we'll have something to say about bots soon.
-
YEEES ;D
-
We're working on it. Hopefully we'll have something to say about bots soon.
Soon as in a few months... I have a lot of other more important (money making) projects on my plate at the moment. Implementing abstract and extensible Ai in a video game is at the bottom of my list (sorry).
-
is it that difficult to make bots ? just some bots who runnign arround and shoot on you if they see you ? they must not have a big C-IQ or ?
-
The main things about bots are:
- they should understand the game mode and behave according to it
- they shouldn't get stuck or be staying in only one area
-> they should either have waypoints and chose their route from them (which basically would mean you'd have to have waypoints for every map you want the bots to work on)
or they should navigate around the map freely without waypoints (even more difficult and most of the times implies they should be 'learning' the map during gameplay or reading out what they learned the last time around)
So the problem of programming an AI comes down to "how smart should they be?". Even a simple dissatisfying AI would have to be taught basic behaviour like "if there are two routes, alwas go left" and recognition of their surroundings (how far will you scan for walls and will you remember walls you already found? -> big scanrange = slowing your system (per bot of course), small scanrage = recogition of walls very late, e.g. if they're already running into them.) making that part non-trivial.
Even with waypoints bots shouldn't unload all their ammo into a wall. Shooting/hitting targets once they know there isn't a wall should be easy, it's all about percentage, i.e. how likely you wan't them to hit.
-
is it that difficult to make bots ? just some bots who runnign arround and shoot on you if they see you ? they must not have a big C-IQ or ?
This is the understatement of the century. Read up on Ai and then come back to me when you are learned.
I have some pretty lofty ideas for the bots and Ai in GE:S. Whether or not it makes it into reality is unknown. Lodle has already made a really great attempt at an offline map navigation tool which would basically rough out a navigation matrix for use with the bots.
Source has a navigation layout editor built right into the engine (most people have never seen it before, but it's there). It is rather rough to use though and requires you to run around the map looking for obvious errors. Either way you can't do navigation "online" it would cripple the game. The bots have to have a predefined movement matrix so that they can do: "I am in facility bottle room and the MWGG is in facility bathroom how do I get there?"
-
Either way you can't do navigation "online" it would cripple the game. The bots have to have a predefined movement matrix so that they can do: "I am in facility bottle room and the MWGG is in facility bathroom how do I get there?"
Where "online" stands for "computing a solution while confronted by the specific task" in comparison to "offline" meaning "computation before the task is even carried out" - for those confused about it.
-
If you make me an R2D2 I will insert it somewhere.
-
1st i dont know what AI should be
2nd : Man i really have to learn english more better.
-
AI stands for artificial intelligence, 'künstliche Intelligenz' in German. Basically you're telling some agent (e.g. a robot) how to behave, how to judge situations or some learning patterns by which he can improve himself. E.g. if it's a robot programmed to walk it would have sensors (for balance, for example) and it would know if it succeeded in what it was trying to do - so it would read out data from the sensors comparing it with the results of its actions and occasionally find a solution to walking on a ramp. This solution's data would be saved, thus meaning the robot would have 'learned' to walk on a ramp without falling over.