Jump to content

AI triggers and thoughts for the future


weta_nz

Recommended Posts

Hi,

There hasn't been a huge amount of discussion about stuff for the future on the Forum lately so I'll bring up one of the subjects that interests me. I would like to see what other people think and/or hopefully even get some comment from boss man Steve

A bit of background: I enjoy single player scenarios, generally smaller ones, which I can finish within a couple of hours. There doesn't seem to be a huge number of these types of scenarios - so I found Quick battles to be a nice alternative with the added bonus I get to choose my forces. However after playing a lot of these there is a 'sameness' to the AI setup and the way the AI responds in quick battles. So I thought I'd have a a go at creating a scenario with a more dynamic AI. Having never tried to make a scenario (for any game) it was with some trepidation that I went into the scenario editor, loaded a quick battle map and started trying to make the AI setup and then react in a more dynamic way that would make sense to me if I was controlling the pixeltruppen. It turned out to be a lot of fun learning how the triggers work and watching how they actually play out in the game. Results ranged from heart warming (Mini ambush with a successful covered retreat to the next line of resistance) to frustrating... but after many hours of testing/experimenting I was able to get some sort of a flexible defense happening with remaining units eventually making a last stand in the village. With more design experience I could get better results and I think working with small scenarios definately puts more pressure on the AI (e.g if you have a limited number of squads and squads get pinned down it affects the 'strategy' quite a bit). A human can can respond to these changes in situation in a much more flexible way then the AI ever will be able to but here are a couple of ideas which may be useful to some scenario designers and may or may not be easy to implement :)

1) The conditional statement - I remember Steve said this may occur for V4.0 which would be great as I think this could be the single biggest improvement for the AI to be more flexible. I imagine a group of reinforcements been able to be sent to the correct part of the map depending on which flank the human player's main effort is coming. E.g If terrain objective 'left flank trigger'is reached by the enemy then the AI defenders reserve group will carry out the orders branch which advances to the left flank OR If terrain objective 'right flank trigger' is reached by the enemy first then an alternative set of orders will mean the defenders reserve group will advance to the right flank. Currently the scenaro designer needs to guess which flank the human player might choose or somewhat artificially channel the player down a certain part of the map. It seems the coding to follow one branch or another could be fairly straight forward but the user interface for setting up Ai triggers would need to change obviously.

2) Delayed or Tiggered support targets. Currently the AI can create a great smoke screen or get the defenders heads ducking for cover with a well aimed barrage at the start of a scenario but there is know way (That I know of) to delay the smoke screen or pre planned barrage occurring. So my suggestion would be to add a delay to support targets. Again this would hopefully be fairly easy addition to the code and would allow the scenario designer to create a more formidible AI attacking scenario as smoke and HE barrages could be delayed to occur with the estimated time it takes to advance to certain parts of the map rather then requiring the AI to actually get eyes on target. An even more flexible solution would be to allow support targets to be triggered. E.g Support targets 'A' and 'B' are designated to be fired upon when a friendly unit reaches an objective trigger. This may be a bit harder to implement though but it would certainly help the 'AI' to execute the attack in a more coordinated fashion - like a human player.

Remember these are just ideas for discussion but it does seem to me that the bulk of the code is already there (with regards to the AI user interface and getting the objectives/triggers to work). So I think it would be great if Battlefront puts some more effort (assuming they aren't already) into expanding upon AI plans/triggers as I think there is huge potential :)

As I said at the start of the post - it would be great to hear other people ideas and/or what Battlefront is planning to do with regards to AI triggers in the future

Link to comment
Share on other sites

Good points Weta.

If friendly reinforcements could arrive only if a terrain or order trigger was activated rather than currently always arrive at a certain time in a certain place it would give designers more tactical flexibility.

I am not sure how easy to code or resource intensive, this would be though.

One thing to make AI testing much easier would be to have the order number and trigger tag name displayed next to the unit UI in 'scenario test mode'. Currently we just have the group number displayed.

Link to comment
Share on other sites

Yes good ideas.

Although the AI will probably never be as good as a human player, there is a huge advantage the scenario designer has over the player: he knows everything about the player's forces. And that compensates quite a bit for lack of situational awareness of a predetermined plan.

What about triggers that hold thresholds for their activation (trigger touch threshold: i.e. 3 tanks, 7 vehicles, 109 infantry)?

Or triggers that have the condition of friendly units as threshold (casualties [%] of AI-group, destroyed [%] of AI-group)?

AI design for scenarios could become tactically interesting in itself.

But with the current interface in the scenario editor, which already is using all available space I see a problem to expand the current system at all. Maybe a scripting language and a compiler would be the easier solution? For the easier plans, the GUI can be used. If a more complex plan is wanted, the scenario designer can write a script in a textfile which is imported and compiled.

Link to comment
Share on other sites

Yes good ideas.

Although the AI will probably never be as good as a human player, there is a huge advantage the scenario designer has over the player: he knows everything about the player's forces. And that compensates quite a bit for lack of situational awareness of a predetermined plan.

What about triggers that hold thresholds for their activation (trigger touch threshold: i.e. 3 tanks, 7 vehicles, 109 infantry)?

Or triggers that have the condition of friendly units as threshold (casualties [%] of AI-group, destroyed [%] of AI-group)?

AI design for scenarios could become tactically interesting in itself.

But with the current interface in the scenario editor, which already is using all available space I see a problem to expand the current system at all. Maybe a scripting language and a compiler would be the easier solution? For the easier plans, the GUI can be used. If a more complex plan is wanted, the scenario designer can write a script in a textfile which is imported and compiled.

Some good ideas in here. Would leave to see the possibility to script similar to the Arma series. But conditional triggers like the ones you suggest would also be very acceptable. At the moment, the designer has to try and think what the player might have done by certain times and trigger the AI accordingly, it would be nice to be able to set triggers for the AI based on what might have happened to their forces, or the player's.

Link to comment
Share on other sites

I agree with Hellas that designing the AI for scenarios can be tactically interesting in itself. When I tried designing an AI plan - it was a lot of fun trying to come up with a AI defensive plan that simulated what i would do if I was in control. The way I would describe the 'fun' is imagine telling your troops at the start of a scenario what they need to do for the whole battle (with some limited desicion making throught the battle) but basically you need to try to figure out the best plan for what you think the enemy might do with enough coverage for some variability in how the enemy might attack. You then hop out of the drivers chair and into the backseat, sit back and watch what happens. It's almost like another game in itself :). The more tools Battlefront can give the scenario designer, the more able the scenario designer will be able to get the AI to react to that variability in a 'human like' manner and therefore the more fun/harder the single player experience will be (without resorting to giving the AI xx% more units as cannon fodder or creating restrictive terrain so the player only has one direction of attack) Note: There are many great scenarios which don't resort to these tactics.

One idea which might be helpful for scenario designers is if you could test an attack AI plan against the defense AI plan - you can then kill two birds with one stone for generating quick battle maps. Also it would be like a game within a game, you could come up with plan for the attack and the defense for a map and see which AI wins the battle! :) I won't hold my breath on that one but I would find it fun!

Link to comment
Share on other sites

Would it work in the current system to have AI artillery assets arrive as reinforcements at certain time windows, and fire at the previously painted support target locations?

I don't know if there's a way for a particular AI asset to be told to fire at a particular location as soon as the asset becomes available.

Link to comment
Share on other sites

Would it work in the current system to have AI artillery assets arrive as reinforcements at certain time windows, and fire at the previously painted support target locations?

I don't know if there's a way for a particular AI asset to be told to fire at a particular location as soon as the asset becomes available.

If Battlefront could add some code to at least add that functionality to the current system - that might be a way for a scenario designer to control where the AI lays down artillery later in a scenario without the AI needing to identify a specific target.

I'm guessing none of the ideas suggested will be new to the brains trust but it would be interesting to know if any of the ideas are on the priority list

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Unfortunately, your content contains terms that we do not allow. Please edit your content to remove the highlighted words below.
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...