OpenArena Message Boards

Several games using the Cube Engine or the Cube 2 Engine use, like open arena, bots with waypoints based artificial intelligence.

These bots use the waypoint file of the map where they are, if this file exists.

If this file doesn't exists, it's different. In open arena, the bots won't work but, in games with the Cube Engine or the Cube 2 Engine, the map will be flooded with waypoint to allow bot use.

It isn't as efficient as a well made waypoint file but, most time, it work well.

So, why not make Open Arena, if the waypoint file is missing, flood the map with waypoint when the first bots is spawn ?

Not really true...
Bots won't move at all when the particular map has no waypoint-file (.wpt) on the original Cube Engine 2: Sauerbraten.
This might be another case on Red Eclipse, but I doubt that.

The big difference between the bot system of OA and Sauerbraten is that bots of OA move more in a pattern, thus more accurate towards the gameplay. However, this makes them rather predictable.
The choice of weapons of bots on Sauerbraten is still pretty dumb; the MG has the most priority, while the GL has the least (if a bot has ammunition for all weapons), no matter the distance between the target and the bot.

Way-points are usually added by the mapper, and hence they are often missing because the mapper could not be bothered to add them (it is a non-trivial, time-consuming task).

Quake3 engines use AAS (Area Awareness System) and it is already automatic, the mapper only has to build the AAS for their map using the "bspc" tool.  Hence it will only be missing if mapper is extremely lazy or map is unfinished.

IIRC, Q3/OA do not use "waypoints", meaning that mappers can't tell bots to follow a specific path (e.g. "push this button, then go upstairs, then enther this door, and finally reach the flag")...
Id tech 3 uses "area awareness system", instead of waypints... BSPC (AAS compiler) divides the map many "areas", and then each bot knows that from a certain area it can reach "n" other areas (I don't know in detail), so in a certain sense, bots decide "on their own" the paths to take. This saves mappers from the need to create many waypoints like in some previous games, although there are some cases where one may miss them!

To somewhow "guide" bots, mappers can place some "item_botroam" entities (giving them a certain "weight" value) around the map, to "suggest" bots to reach certain places where they seem not interested do go by their own (bots are attracted by items, and may not be interested by areas without items... placing item_botroam there can help bots explore the whole map.). Also, "info_camp" entity allows to suggest to bots places adapt to "camp" (stop there and look around, waiting for someone. Some bots do have a more "campy" personality than others.). It is possible to "suggest" bots do to not enter an area by creating brushes with "common/donotenter" shader, or to physically block them using "common/botclip" brushes. However, you cannot specify a real "waypoint path"... bots have to discover by their own that button A opens door B to reach room C, and sometimes this may be a bit uncomfortable... but that's the way Q3 worked.

After you compile a map, you have to run BSPC command to create its .aas file, which allows the game to know where bots can go. Sometimes the map is too complicated and it fails, or sometimes mappers forget to include the file.

See also: (DO NOT LINK) h t t p s : / / openarena . wikia . com/wiki/Mapping_manual/Bot_play

I understand better how it work, now.

I believed it was like in Sauerbraten.

I can relate about the whole waypoint thing. I've experienced the "no waypoints" problem so many times in the Unreal Tournament games that I've lost the count on how many maps lacked these points because mappers never bothered to add them.