Kyall asks “Is RTS-style AI being currently used in FPS games? Because I know that RTS-style AI is multi-layered and complex to the point of creating a strong strategic opponent, but is this same style of AI being used in FPS and if not, why not?”
You’re right that first-person shooters as a genre are making slow progress in terms of tactics, and indeed AI techniques from real-time strategy games would help tremendously. But it’s important to keep the following in mind:
Story-driven action games don’t need much real-time strategy.
Some tactical terrain analysis is already used in shooters.
Multiplayer combinations of FPS and RTS avoid the need for AI.
AI techniques from RTS are useful but not easy to integrate!
I’ll explain to you what I mean using a few examples…
Screenshot 1: Field Ops — A Real-time Strategy Shooter
Let’s face it, most FPS these days (like Halo, Half-life) are very linear. Even if there are a few open areas or multiple paths possible, they are limited and planned to join up.
Because of this design choice, level designers effectively fill the role of the high-level AI in real-time strategy games. Level designers:
Are responsible for placing enemies strategically.
Must balance the difficulty settings for each level.
Implement scripts to handle special situations.
Using a combination of design during development and runtime scripts, it’s possible to simulate most “tactical” decisions without any kind of AI — which works fine for these story-driven games.
Tactical AI in FPS Games
While tactical AI is not always necessary, it does improve the gameplay and reduce the workload on level designers. Some first-person shooters already include a certain amount of tactical AI, notably for terrain analysis and squad behaviors.
- A. Tactical Movement
In games like Halo and Killzone, paths are computed with a custom heuristic. This means that combat factors are included when selecting the best path:
Is the actor moving through the cone of fire of a teammate?
Is the actor exposed to the player’s weapons?
The typical shortest-path algorithms can be easily modified to use these extra factors, but it can be a bit slower to perform the search.
- B. Cover Location Selection
Most FPS games store cover locations for the actors to use during combat. These are either computed automatically or annotated by developers. At runtime, the actors select these points based on a wide range of tactical factors, such as:
The distance to the cover location from the current point.
Does the path into cover go too close to the player, or expose the actor to fire?
The position of the cover point relative to other actors (for better team distribution).
The position relative to the player, so flanking is encouraged.
These factors are used in many games already, but it’s rare to see these behaviors shine when the actors die so quickly.
- C. Choke Point Analysis
- Most games use some kind of navigation graph that expresses connections between rooms and corridors. Using this information, it’s possible to identify parts of the graph that are “choke” points where the player must go through. Half-life for example uses this information to help improve NPC placement and movement during combat.
Now it’s true that much of the logic for FPS enemy AI is about filtering out the stupid decisions, but nevertheless, you could classify the result as tactical decision-making!
Screenshot 2: Zombie Master — The player under siege by a horde.
Blurring the Lines
Shooters are increasingly aiming towards RTS levels of strategy. For example:
Hidden & Dangerous was one of the first squad-based tactical shooters with a first-person view and a top down map too.
Full Spectrum Warrior achieved this on the console platform with an intuitive interface for ordering squad members around.
Field Ops plays off fist-person fighters against a RTS player. Same level, different view.
Zombie Master does the same as a Half-life 2 mod facing off first-person players with top-down controlled zombies.
Iron Grip is another Half-life 2 mod combining RTS and FPS in a world war setting.
These games were designed to require a level of tactical AI. Now it can be disappointing at times, but that only emphasizes the challenges of getting this right. Games with a more open combat areas like Battlefield and Enemy Territory are even more of a challenge for the AI, which explains why the developers focus on the multi-player aspects of the game.
RTS AI Applicable to Action Games
Going forward, in FPS games with open combat areas, more units, and less emphasis on linear stories, RTS games may provide the solution for better AI. Notably:
Influence maps help tactical reasoning by providing information about dominance of teams per area.
Commander AI to determine what kinds of units are available and where to send them at runtime.
The challenge for programmers is integrating these techniques with FPS AI. This isn’t as easy as it sounds; it’s hard to implement AI behaviors that take into account all tactical factors, but game studios are getting better at it with each new game.