Advertisement

Finding cover - how to implement?

Started by October 27, 2000 02:34 PM
4 comments, last by Tom 24 years, 1 month ago
Anybody notice how the bots in Unreal Tournament are particularly good at ducking for cover? I''ve actually seen bots duck behind a separation between two openings in a wall, which is exactly the type of response you''d expect from a thinking creature. One of the greatest goals in contemporary game design is to create impressive AI to alleviate the need for Internet and LAN gaming. (Few people have access to a LAN, and Internet lag can ruin the multi-player experience.) In my quest for better AI, here is my question: How do I get bots to recognize cover? This applies to walls (which are part of the game world) and obstacles like barrels, crates, or even other players. I don''t expect a complete answer to this question, but some leads would be great. I''m looking for popular methods, so if anyone is familiar with AI coding techniques used in Unreal, please let me know. Thanks.

GDNet+. It's only $5 a month. You know you want it.

I''ve seen it done where it uses the draw mechanism to find an object nearby that causes the player to be occulted. If you tie this in with an algorithm that tells it to only search within x distance. It should be relatively fast.
Advertisement
If it knows where the player is/will be, then it can guess where the player can/cannot see. Simply move so that an object is between the player and itself. If it doesn''t know where the player is, then it can find someplace where very little area is exposed or something.

Think outside the octahedron


I am so glad that Landfish is finally getting the respect he deserves.
Several billion trillion tons of superhot exploding hydrogen nuclei rose slowly above the horizon and managed to look small, cold and slightly damp.-The Hitchhiker's Guide to the Galaxy by Douglas Adams
Since your rendering engine is going to select the surfaces which the player can see, simply selecting the nearest visible object that is big enough should be sufficient. However this would mean your4 rendering engine would need mark the visible object objects. This is not such a bad overhead and if used correctly can even reduce rendering time.

Illusion is meaningless, unless it affects reality.
Illusion is meaningless, unless it affects reality.
I heard someone that said he made his pathfinding system value ground more it it wasn''t in the player''s sight. So his bots would try to get behind the player.
Hmm... Perhaps as you are drawing the scene, you could mark the leaves (I assume you''re rendering from a BSP) that are drawn as being "visible" to the player and use that information at a relatively low cost.
If a man is talking in the forest, and there is no woman there to hear him, is he still wrong?

This topic is closed to new replies.

Advertisement