Biters pathing logic

[Guilwin]

Hello engineers!

I have been trying a different approach at building defenses than what I did before which was "put turrets along the entire wall defense".
I am purposedly leaving parts of the wall open to funnel biters and kill them there.
However, it seems they are completely ignoring these openings and just chewing at the wall as soon as they come close to it.
See and example with my base below, biters just tried to go through the wall at the blue arrow, despite the big opening in the wall very close.

wall.png (875.76 KiB) Viewed 5527 times

This is not the only example I have seen of biters doing that - it looks like they almost never go through my funnels - but this was the one where they attacked the wall the closest to an opening.

What is the logic followed by the biters when pathfinding?
Do they just ignore walls when calculating their path and bite their way through it everytime?
Is there a maximum distance they will deviate from a direct path before they start just chewing at the wall in their way?

My goal was to make for a less expensive defense wall by building regularly spaced funnels with high concentration of turrets and big openings to lure the biters, but it does not seem to work at all. If the "maximum deviation" is so small that they will go through the wall in the example above, I will not be saving as much as I thougth I would, if this technique can even work.

Has anybody tried a similar approach before?
Are there any forum posts/articles about biters pathing logic that would help me understand their behavior here?

Cheers!

[netmand]

you can't use walls to funnel biters. only cliffs and water cause them to alter their path.

the bugs have very little tolerance for destruct-able objects in their way (i.e. trees, rocks, walls)

[coppercoil]

Walls won't work, bitter pathfinding just ignores them.

Of course, linear defense is ineffective. But you don't need directing walls for concentrated defense. Biters always switches into attack if there's a turret nearby. So you need to distribute turrets by double switching radius (minus some margin for placing errors). This is approx. 2.5 of turret shoot radius:

ConcentratedDefense.jpg (37.88 KiB) Viewed 5490 times

You don't need to cover gaps because biters will come for turrets

[Guilwin]

Thanks for the replies.

So does it mean building a closed wall is exactly the same as building a wall with openings in front of turrets?
I was thinking biters would path towards the turrets more easily if there is an opening.

I see how a defense like the one in the picture you posted would work perfectly well early, but I am trying to design an optimized defense wall for when biters come in such big waves that they would easily overwhelm a couple of isolated turrets. In my current base, I just have a continuous row of laser turrets all along the (100% closed) wall, even a double row in the most attacked zones.
I was thinking about building regularly spaced funnels like the one below.

wall 3.png (303.28 KiB) Viewed 5427 times

Now I understand that the maximum spacing will be about 2.5 gun turret range and the funneling will not even work that well.
And that I should probably close the wall completely.

[MEOWMI]

Biters will typically only ignore walls if there's a turret within shooting range and the wall isn't directly in the way. This means you still need more or less a solid line of turrets at the border, but biters are willing to take a slightly longer path. Using this fact, many players create additional walls with gaps in them to lengthen the path biters have to take. The best designs I've seen involve funneling them through one choke, which also accomplishes the former but allows for flamethrower turrets to deal even more damage with their AoE, but even in these cases, it's not one massive choke covering a wide area as this aggro pathing mechanic more or less limits the coverage of such a choke to the range of a turret. Here's one specific example I remember seeing: https://www.reddit.com/r/factorio/comme ... e_picture/

[netmand]

An interesting behavior I've seen is that biters progress really slows when coming through trees. However, if there are enough to clog them all up between trees some of them will start taking the trees out. Nevertheless you can take advantage of the slowed progress to mow them all down.

If you don't have a convenient tree-line you can use wall pieces spaced out enough to have small gaps that they can go through so the walls aren't immediately damaged.

I've also seen people use express belts to influence biters off their path, especially to keep spitters out of range.

[coppercoil]

By the way, there’s one important question about funnel width. IMO funnels that are posted in this topic are way too wide. It will not reduce bitter flow (bitters per second arrived - bullets per second spent), so you cannot reduce turret count. Single line of turrets may be not enough. Funneling has no other benefits; turrets can shoot together in any direction if you pack them tightly.

Funneling may be practical if biters are stooped and forced to arrive one after another (less bullets per second – less turrets required). Though, it may be difficult to design to not damage walls (unless you have deployed construction bot network for repair). I tried but I didn’t succeed. Interesting design pointed by MEOWMI looks like… a hack, that works just because bitter’s patience was tested very carefully. I think more tests should be made to be really sure whether it's sustainable. And… is it cheap?

[torne]

You don't have to reduce flow for there to be a benefit, but you're right that the designs posted here won't really help. If you can convince the biters to take a longer path, even if that path has no bottlenecks and the overall flow rate remains the same, you can keep them inside turret ranges for longer than if they ran directly toward them, and that helps a lot if you have flamethrower turrets (because they're AOE) and also helps when there are larger biters in the mix (as long as the wave isn't *all* the larger ones and there are some smaller ones to reduce the average number of shots required)

[coppercoil]

If biters run in zigzag pattern (though it may be difficult to design walls that would work for any direction without being eaten), its time under the fire increase, so we can also say that turet shooting range virtually becomes larger. That's good. How good is it? A gain is equal to General DPS * Shooting range increase / biter run speed.

But IMO zigzags has no real benefits for flamethrowers, because straight run through the flames also kill very effectively.