Troop AI

[RottPhiler]

1. Instead of using Euclidean distance to determine the nearest building, use your own algorithm that determines the best path to reach said building to determine the nearest building along that path.

Often times my tanks will be closest (in Euclidean distance) to a tower or mortar behind a wall and will happily roll in front of an anti-tank gun to shoot it. Also, cannons and shooters will determine that a mill is closest (in Euclidean distance) and happily walk in front of a redoubt or tower.

2. Rally: weight your algorithm by wall strength to determine shortest path based on quickest time of travel and breaking necessary wall segments to get there.

You must be doing some form of this already, but currently two things happen - a) multiple units attack multiple wall segments to break through, and b) units go the long way to attack a partially broken wall segment when breaking an intact wall segment closer to the rally point might be quicker, and therefore the shortest path.

Disclaimer: this could break something else though, so extensive testing might be required. It will help if you could recreate the scenarios above.

 

[Delta3519]

Yes for real. I 100% agree with you

 

[Tsamu]

1. I think we have disagreed about this elsewhere. My "assinine" reason for not wanting a change to the AI is that players have built "traps" in their bases (especially war bases) to exploit the current behavior. I repeat my dissent here to give participants in this thread another consideration.

2. Completely agree with you on this suggestion. Watching an archery beat on a wall for 2 minutes is very frustrating.

 

[RottPhiler]

For 1. I feel like either I'm not understanding what you mean, or you're not understanding what I'm trying to express. Why I say that is because, I think it is impossible for you to build your traps considering this faulty AI behavior (and also to a minor point: rebuilding your traps to suit the proposed new AI behavior should not take away from any of of your genius trapping abilities).

If you're talking about leaving certain holes open in your walls for enemy troops to walk through them, and consequently get spiked, I don't see how that affects you at all. Troops will still walk through the exact same trap.

My point in 1 is that troops choose to attack a building, and then walk a certain path. I didn't propose to change this certain path at all. Let the troops walk the exact same path as they previously would have. I only propose attacking a different building (one along this path), instead of the one they originally targeted. How does that change your trap locations?

I'll try to create a visual example for you below

....Tower
B
....WWWWWWWW
....W
....W Mill
....W
A..W

W is a wall segment. The troops arrive at point A. The mill is the closest target in Euclidean distance, and therefore they lock in on A, and target it, except they move to point B to position themselves to attack the mill. So the AI already knows that it will move from point A to point B to attack the mill, and will do so irrespective of where the traps are. All I'm asking for is that once it makes the calculation that it will have to move from point A to point B to attack the mill, then it should consider that at point B, the tower is closest in Euclidean distance than the mill, and therefore it will have to be attacked first.

So essentially at point A, the troops have the following choices
a) Stay at A and attack the mill without moving to point B
b) or determine that to attack A, it will have to move to point B, move to point B, and realize that the closest target is the tower, and not the mill.

How does this change affect your trap placement?

 

[Master Contrail Program]

Bump this too. The targeting and pathfinding(especially for howitzers) is absolutely atrocious and needs to be addressed.

 

[Kaiser Shag]

I don't remember that I've ever seen a reponse from Nexon concerning this terrible IA problems, TinSoldier are the dev's actually working on this point or is it totally ignored ?