Tricks to Help the AI?

[SFJaykey]

Just wondering how you experienced scenario designers manipulate things like map design, setup locations, victory conditions, etc. to guide the AI on attack and defense. I've certainly seen very strange AI behavior, but also seen it act reasonably and am sure scenario design figures in this heavily.

Thanks,

[JasonC]

Avoid completely hidden defenses. The AI orients on its sighting reports, so it needs to see things early. A completely hidden defense will sometimes leave the AI sitting at the start line doing nothing. Instead, have a forward outpost platoon or a few vehicles in locations that can be spotted pretty early. Provide small flags at "intermediary" positions. The AI also focuses on flags - particularly enemy held ones.

Avoid maps that have only a single covered route to an objective. The AI will string out its whole force along that route and is easily defeated. Multiple covered routes and seperated forces at set up ("converging" lines) can avoid this "bunch up" tendency.

It is much easier to get the AI to defend credibly than to attack credibly. It is much easier to get the AI to attack with significant amounts of armor in its force mix than with mostly infantry, unless the terrain is quite heavily forested. In heavy forest, the problem is spotting any enemy.

The AI has a poor understanding of methods of tackling thicker front armor AFVs. Avoid giving the human player a strong advantage in that respect. It is challenging and fun to take out thicker AI controlled AFVs. Even quality armor battles also work, particularly when the AI ha greater numbers.

When attacking, the AI needs odds well beyond what humans get by default. 2:1 to 3:1 point totals are the right range. In addition, expect the AIs artillery to be less then fully effective, and concentrate on the more rapidly responding modules. So count its arty points as half their usual value.

It does not plan far ahead with arty fire, tends to fire a whole module at single targets, and often fires smoke when it shouldn't. You can also boost the experience level of FOs to reduce reaction time. Air support is a wild card but the AI can't mess it up "on purpose".

Reinforcements arriving in modest size groups can "pace" the AI through a battle. Instead of giving it all its forces at once and expecting it to decide how to allocate them to areas of the map, think through some of those issues yourself and add reinforcements behind particular sectors at different times. The new threat created by unexpected forces from a different area, or a second wave behind a first, makes for exciting play.

The AI handles flag dominated victory conditions better than exit based ones, if it is the one exiting. Forcing a human player to exit through the AI is also a decent VC, particularly if the map sets up a few "chokepoints" and the AI sets up near those ("roadblock" fashion). ID such positions with flags so the AI will know they are important.

Put flags on tactically important bits of terrain, not ones meant to represent some higher echelon thing unrelated to the fight on the map. Not a road junction e.g., but a hill top or a large heavy building or large body of woods. The AI tends to puts troops of all types all around flags. This should make sense instead of being a trap.

It is challenging to confront the human player with situations that require getting the most out of an element of the combined arms package, facing types especially good at countering them. E.g. Pioneers needed for AT ability in a village, or ATGs must stop large numbers of tanks in relatively open country. Not scads of HMGs against infantry in the open, or StuGs against T-34s at long range.

The latter are good things to give the AI to make it stronger. But with some "out", like on map mortars and FOs in the first case (rather than too easy tanks), and decent cover for approach runs in the second.

I hope these help.

[flamingknives]

Setting starting positions and either locking them or using scenario default settings can help.

Victory Locations (VLs) - the flags - can be used to create an attack route for the AI too. Trouble is, then the human player can work it out too.

Alternatively, you can give the AI an advantage, such as experience levels, simple numerical superiority,giving the Human player an unbalenced force (too few ATGs/armour/infantry etc.), or setting up terrain that doesn't suit the humans force mix (Panthers/SPGs/PanzerJagers in close country, for example)

[SFJaykey]

Thanks guys!

[British Tommy]

Excellent bit of work there Jason! I saved the text and will use it to help me plan my next scenario.

Thanks mate!

[manchildstein (ii)]

if you want the ai to stay where you set them up, create an exit scenario where the ai defends with no flags... this seems to keep the ai where you left it... no bizarre counterattacks against flags... learnt this from the 'road to smolensk' scenario...

there was another scenario, 'wittmann's cross of iron'... where the ownership of the east and west mapboard edges was reversed... i think this somehow caused the soviet tanks to attack the germans...

you can get some spectacular turn 1 offboard artillery barrages if you can get human-controlled units within the LOS of ai-controlled units at the start... time and time again i've seen the ai hit with lots of artillery if there are human units within LOS.

remember that, with borg spotting all the ai has to do is to is have LOS to the human-player's units and the ai can then dump beaucoup artillery rounds there...

sometimes it is interesting to put ai-controlled units behind the human player's at-start positions... this would represent ai-controlled units which remained hidden and are now ambushing from behind...

you can sprinkle the map with ai-controlled sharpshooters, lmgs, tank hunter teams and such... one favorite is to used split smg-squads and make nasty defensive belts in wooded or otherwise low-los areas... split smg squads because all you need is 4 or 5 guys to spring a nice ambush... then you have twice as many units to cover ground with...