Battle AI Blog

[Fisherking]

Tuesday Nov 10, 2009

Battle AI - By Mike Simpson

Just before the end of Empire the lead Battle AI programmer left CA to return to his family up north.
Unfortunately, thanks to Mr Wilberforce’s efforts 200 years ago we couldn’t stop him. It left us with a battle AI, which at that stage, struggled to beat good players in a fair fight, and was pretty much at the mercy of great players, even with a level of handicap (I call it cheating) that is all too obvious.

Since then we’ve had our most talented programmers pick up where the deserter left off, but becoming the master of a chunk of code that took almost three years to write is not an instant thing. In the updates so far we’ve got rid of some of the worst behaviours that are close to the surface, and have started to tackle deeper issues like unstable decision loops that cause the AI to mill around rather than hold its line and fire. We’re also starting to add new behaviours, for example taking better advantage of hilly terrain. These improvements take the code further than we’ve been able before and will be there for Napoleon but we’re not sure yet whether we’ll be able to reverse them back in to Empire in a future update - the code has moved on. If we can we will.

Our overall aim with the improvements is first to get rid of any erratic behaviour, second to improve general performance to the point where the obvious handicaps can be removed, and then to add a greater variety of ’smart’ behaviours. None of these have a fixed finish line - it’s a process of continual improvement, and each game will get AI better than the last one. Including Napoleon.

Progress is frustratingly slow but thankfully rewriting the Rome: Total War codebase has left us with a clean codebase that is easier to work on, and an architecture that has way more potential than Rome’s. The main difference is the shift to a goal oriented planning system rather than a static system that has no long term plans. This has yet to fully pay off. But it will. When it does I’ll talk about it again.

Battle AI is not rocket science - its way, way harder than that. It’s so difficult that very few strategy games attempt it. Most use simple scripts of canned behaviour that fire when you bump in to them, and very simple swarming behaviours. They’re limited, and are “gamey” rather than real world. What I mean by that is that the tactics you use to beat them are something that you have to learn for each game or sometimes each scenario/level, and bear no relation to reality. What we strive for is a game where real world tactics actually work. It’s not the easiest path to take, but it is the most rewarding.

If you’re a genius C++ programmer, you understand exactly how difficult this problem is, but still think it’s the most interesting code problem in the world, apply for a job. We’ll find space for you.
Posted by Mark in CA Updates on 2:42:07PM Nov 10, 2009

http://blogs.sega.com/totalwar/2009/...-mike-simpson/

[Owen Glyndwr]

That's nice, now where are the modding tools?

[gollum]

This was a nice entry - unfortunately the main programmer leaving in the midst of development happens in many games (it has happened to Crusader Kings of Paradox too) - and its a thing that very little can be done about. Its to be commended uopn that a single coder was in control of the whole, this gives more cohesion and allows greater amount of control and detail over the code due to a single person being able to achieve greater knowledge over it, rather than compartmentalise the effort in between other coders and introduce the "broken telephone" effect of inter-team communications.

The battle AI of Empire TW is by no means finished and i still feel cheated that Napoleon won't upgrade it - but it undoubtedly has potential to be better and the problem at hand (coding battles that all units have missile weapons) isnt an easy one indeed.

PS To all those that have argued that M2TW had a better battle AI than RTW:BI, note that it is again stated in this entry that the best AI battle code they had was in BI and not in Kingdoms; apparently only peripherals were touched in M2TW by CA oz; it seems that there was no or very little work on the AI for that game, so statements from CA in general and Mr Simpson in particular at the time, that said "we are improving things accross the board" (the TW Story) were most likely simply marketing talk. As many have said for M2TW it was little more than an extensive mod for RTW.

[Peasant Phill]

Since then we’ve had our most talented programmers pick up where the deserter left off

I hope he means this humourously, otherwise we're playing the blame game again.

[Martok]

That's indeed most unfortunate. Especially the part about the upgraded battle AI not being integrated into Empire.

I hope he means this humourously, otherwise we're playing the blame game again.

I concur. To be fair, I *do* believe he said that with his tongue planted firmly in his cheek, but he probably should've placed quotes around it ("deserter") or otherwise worded it better to reduce ambiguity.

[antisocialmunky]

That's some good insight. Never thought about that happening but I guess those are the risks you take when you hire a specialist like that. Maybe NTW will have some of the bugs worked out...

I do disagree about the 'canned reactions' part. All game AIs use a deductive process to narrow down a set of possible actions to a single action that the AI has been scripted to do. It doesn't make them up as they go. I guess if you had some sort of crazy evolutionary algorithm for an adaptive AI like everyone wants to see it would be different but ETW is just as canned as everything else.

At the end of the day, I rather see CA produce a game as great as they market it than another total failure of a game.

[peacemaker]

will be there for Napoleon but we’re not sure yet whether we’ll be able to reverse them back in to Empire in a future update - the code has moved on. If we can we will.

Heh. I knew it. Try to convince people to buy NTW because of the 'fixed' BAI which UNFORTUNATELY can't make it into empire.

we’re not sure yet whether we’ll be able to reverse them back in to Empire in a future update - the code has moved on. If we can we will.


None of these have a fixed finish line - it’s a process of continual improvement, and each game will get AI better than the last one. Including Napoleon.

Seems a little contradictory, no?

Since then we’ve had our most talented programmers pick up where the deserter left off

The second they mentioned the lead programmer leaving, I knew they were playing the blame game again. CA has proved to entertain me though nowadays...

[Zenicetus]

CA could make an immediate, major improvement to the BAI just by eliminating the ability for the AI and the player to garrison buildings. In a more realistic game, both the AI and the player could choose to garrison any building or structure on the battlefield, which would keep the army together. Instead, (and probably due to 3D graphics limitations), only a few scattered buildings can be garrisoned, and they're never close together. The player is smart enough to choose one building to put troops in, with supporting troops nearby, using surrounding buildings to protect the flanks. The AI never does that. It just occupied all possible buildings with single units, while the rest of the army remains somewhere off in the distance. This allows the human player to pick off units one by one. Even when driven out of a building by artillery fire, the AI unit never rejoins the rest of the army, and the rest of the army never provides support. They're sitting ducks.

I know there's a lot more that could be improved in the BAI, and this doesn't happen all the time because not all battles included buildings that can be occupied. But this is such a failure of a game "feature" that they should really just remove it. Sometimes simplifying the design is the right answer to improving it, instead of piling on more features that just become player exploits. I doubt that CA would ever do this, but it would immediately improve at least some of the battles.

[Claudio]

Rubbish. Nothing more than an attempt at raising sales for Napoleon.
I'll be proven right as soon as Napoleon is released, which I won't be buying, by the way.

[Crazed Rabbit]

Makes me wonder; what would CA do to cause a very important programmer to leave the project shortly before launch?

Still, I hope they make a competent AI and I hope even more they put it into ETW.

CR

[Beskar]

Makes me wonder; what would CA do to cause a very important programmer to leave the project shortly before launch?

Still, I hope they make a competent AI and I hope even more they put it into ETW.

CR

Probably because he didn't have the time to finish it, and SEGA executives wearing black suits, sunglasses and earpieces dragged him from his chair and took him to a black professional car and he was never heard of again.

He was later found in the basement of EA, made to suffer scorn for eternity, programming DRM.

[A1_Unit]

Well, this new system of AI has me optimistic - the less scripted behavior the better.

[Beskar]

Well, this new system of AI has me optimistic - the less scripted behavior the better.

Being honest, scripted behaviour actually makes it better. less scripted can usually overcomplicate or overwhelm.