M28AI Devlog (v165)

maudlin27

AI will upgrade all mexes from T2 -> T3 at once, causing a massive and persistent mass crash. Could be reclaim related? Eg I collect a bunch of reclaim, the AI qu

I think this was your teammate the first time - M28's logs for upgrading a unit didn't trigger when 4 t2 mexes were started at your teammate's base.
For your mexes, the issue is as a whole your team had lots of T3 mexes, and lots of mass stored (as your teammate had about 3k mass while you had none) - so M28 started upgrading multiple T3 mexes on the basis the mass from your teammate could fund it.
I'll tweak it in the next version so in MOBA mode M28's team resource sharing logic is enabled (which should largely mitigate such issues). It'll still be a problem in non-MOBA teamgames with shared armies though (where a teammate with lots of mass and/or T3 mexes will result in too many of your mexes being upgraded if the teammate sits on the mass instead of sharing). I can't see an easy solution to that though, as in many cases people wouldn't want the 'share resources' team logic to trigger.

Re the T2 shields being destroyed this looks like a problem with its backup shielding logic where a shield that is 99% complete is being completed (when that should only happen when under enemy artillery fire), I'll see if I can fix for the next update.

maudlin27

v30 Update

24 misc changes and fixes, including:

Improved kiting for frigates/navy generally vs enemy hover and when bombarding shorelines
Fixed several issues with shared armies where certain units (novax, fatboy) wouldn't benefit from the inherit status game option
Mobile shields should be less likely to suicide into enemy units
Asfs should be less likely to suicide into enemy air units at the enemy base
Fixed a bug for SMD where M28 would always think the enemy had a nuke launcher (and so build a pre-emptive SMD too soon)
4 LOUD specific changes, including reducing one of the MML builders where the enemy has TMD

Acknowledgements

C04spoon – Identifying and providing a fix for a case where T2 arti shots could miss M28 units in LOUD
Vortex – LOUD replay
PurpleMossClump – LOUD replay
Azraeelian Angel – LOUD replay
RD – various LOUD replays
Sioloc – replay where M28 would destroy and rebuild the same shield
Innomen – Highlighting that inherit army status wouldn’t work for novaxes and fatboys

MrPhalanx

Heyyo Maudin. Thanks to also release the M28 for Steam , but I encountered with one problem . I'm using the Steam FA, with various mods of course + Sorian AI as well, but the UEF T2 Engineering Suite (The Kennel), when built, messes the bottom interface, you can select it, but actually the interface is missing, and I cannot upgrade it, checked the T2 Cybran E.S. The hive, there was no problems with it, neither with the T3 Kennel, only with the T2 one. And it was working properly until I installed the M28. Can you look for some solutions for me ? Cheers: Mr.P. PS: I'm using the latest M28 for Steam what I found on ModDB.

MrPhalanx

Also the AEON T4 CZAR unit seems to be broken with M28 , because when it finished, opens up and start firing to the ground, because of this, its utterly vulnerable to air and seems idiotic. Also it doesn't seem like to use units from Xtreme Wars, what is my ALL TIME favourite...

maudlin27

@mrphalanx I'll look into the kennel unit to see if I can reproduce and fix the issue for the next release. In respect of the Czar I'm not clear on what the issue is from your description - I'll run a sandbox test to see if there's anything obvious in how it attacks units, but I'd have expected it to be vulnerable to enemy air units (as it's primarily an anti-ground unit)

MrPhalanx

@maudlin27 Thanks for the answer mate. The problem with the CZAR is that is the AI automatically puts it into planetary annihilator mode (that means when the carrier opens up) right after it built it. And when in planetary annihilator mode, it cannot attack aircraft, the Sorian ai only opens it up when reached an enemy base/ cluster of units...

MrPhalanx

And I know maybe I'm asking too much, but can't you make at least a M28 -Experimental, what can focus on even T1 Experimentals from the start ? Because now the AI only after reaching T3 starts to build T1 Experimentals slowly (mostly from Total Mayhem, but there is plenty T1 in Xtreme Wars too), almost none of T2 Experimentals, and a ton of T3 Experimentals. Sorian AI does this part mostly properly, the only problem with it that sometimes it even rushes to build a T1 Exp, even to an extent to it forgets to build a proper economy/minimal base defenses.. Your AI is way better at this, but it does a little slowly, okay I did not use the AIX, just the AI(Turtle, Rush, Etc...).

MrPhalanx

I want to make clear the CZAR problem, I looked into it, when it opens up it still can fire an air unit, that's not a problem, the problem is, when your AI builds it, it AUTOMATICALLY FORCE ATTACK GROUND, so because of this, the unit cannot attack anything other land. Sorian AI has no problems using it properly.

maudlin27

I've not been able to reproduce the Czar issue you mention (even if I activate the total mayhem mod since you mentioned that in your reply), but from the description it sounds like it's broadly working as intended since it's meant to be used to attack ground units rather than air units. So while it might not be optimal (re there being some scenarios where its AA abilities would be good), the difficulty with reproducing it coupled with then the difficulty in having logic that figures out if it wants to have its air to ground attack active vs its AA attack active means it's unlikely to change.

I've fixed the bug with the kennel not upgrading though (will be in the next release).

M28 should build T1 experimental base defences, but I've no plans to have it build other T1 experimental units (e.g. tanks etc.) as M28 is setup on the assumption that engineers can only access experimentals at the T3/experimental stage (not T1-T2) so it'd be non-trivialto add and I'm not clear that it'd have a significant competitive benefit that would justify the extensive recoding required (compared with just getting T2 land or more T1 non-experimental units).

maudlin27

v131 Update
28 changes and fixes, including:

8 adjustments for when M28 hits the unit cap, primarily focused on when ctrl-king its T2 units is not enough
Reduced M28's tendancy to overbuild aircraft carriers
Engineers should no longer suicide into enemies firing at them that it lacks radar coverage of
9 LOUD specific changes (including adding a workaround to allow prioritisation of fatboys) and 1 steam change (fixing a bug with T2 kennels not being upgradable)

Acknowledgements:

MrPhalanx – Mentioning T2 kennels couldn’t be upgraded in steam
RD – Bitterwind Crag, Elephant Rock and Mellow Shallows replays against v1.29
Vortex – Replay vs 1.30

maudlin27

v132 Update
21 changes, including:

Gunships should group together better, and take into account how successful they've been in the game with engagements when deciding whether to attack
Fixed some issues with transports that would e.g. prevent them dropping early when dying to inties
Reduced tendancy to build more factories late-game
10 LOUD related changes, including reviewing/tweaking various LOUD focused PRs from Azraeel (covering building more T3 PD, bombers, ecoing more, and changes to unit prioritisations)

Acknowledgements:

Azraeel - Various pull requests

MrPhalanx

Thanks anyway, I check the Ai again ..

maudlin27

v133 Hotfix - Summary
11 Changes and fixes, including:

M28 should no longer try and protect a friendly teammate's core base afer the teammate has been killed and their base destroyed
Changed how M28 prioitises which zones to request units/give them orders, to take into account enemy buildings (in theory making it slightly more likely to attack undefended enemy mexes over defending a friendly location)
Fixed a bug introduced in the last version for managing gunships that could sometimes break their logic
7 LOUD specific changes, including supporting factory enhancements, and updating the ACU threat assessment to take into account the ACU health and to account for the massively higher mass cost given to it in the blueprint (10k, which is almost 200 t1 tanks)

Acknowledgements

Vortex - several LOUD replays

maudlin27

v134 Update
Another 13 changes, including:

Greater focus on ecoing, with mexes being more likely to upgrade if held for a long time, while making it more likely mexes are prioritised over factories if stalling mass
ACU should be more likely to get high value reclaim near it if it's not in immediate danger, even if it would like to get a gun upgrade
Slightly reduced the amount of engineers to be built in cases where the zone has run out of assignments to give them relatively recently
4 LOUD specific changes, including fixing a bug that would result in lots of land factories upgrading at once, and reducing MML production where the enemy has TMD.

Acknowledgements
Vortex - A number of LOUD replays

v135 Update
Re-uploaded as I forgot to turn off some debugging options.

maudlin27

v136 Update
18 changes and fixes including:

Disabled some of M28's kiting logic if it has poor intel
Improved experimental targeting when faced with an enemy under shields (so they should focus down the shields first)
Fixed a bug that meant submarines wouldn't run from torpedo bombers
6 LOUD specific changes (Aeon land experimental preferences, and various gunship adjustments)

Acknowledgements

Vortex - several replays
Azraeel - Replay that highlighted the bug with subs not retreating
Kemi - Replay showing overbuilding of protector bots in LOUD
Waffelznoob - Replays of M28 being crushed, and crushing

v137 - Hotfix for a LOUD specific bug

Saver

Hello everyone, I have a little thought for amphibious units that move underwater. I can cite the GC as an example. When the AI uses this unit, I have often found that the unit will stall in the water on banks if enemies are too close to the water. The problem, however, is that submarines or other enemies with torpedoes also attack them and the GC makes no real attempts to save itself. Or rather take up the fight on land, where he also has weapons.

Unfortunately I have no idea whether this can be fixed. But oh man, what hasn't the M28 already accomplished

maudlin27

@saver Anything like that I’d need a replay to assess as there will be competing logic:
-Retreat to avoid damage from subs
-Avoid a cycle of retreating and advancing all while taking damage without doing anything
-Avoid making landfall to die to a superior force
-Avoid overreacting to a small naval threat

The logic will likely have room for improvement in specific scenarios

Saver

@maudlin27 Hi and thank you for your quick response. Unfortunately I don't have a replay right now. However, I could see it quite often in map like seton. If I have a good example, I'll be happy to pass it on. However, I play with my unit mod. and hope it's okay for you too.

maudlin27

v138 Update
13 changes, including:

Units should try and take evasive action from a nuke missile they can see approaching them
Fixed an issue that could cause far too many land scouts to be built
Adjusted when M28 builds factories and power generators to try and reduce cases where it can overflow mass
Fixed an issue with if M28 was given control of allies in Fort Clarke Assault it would take control of the starting Ahwassas once the cutscene ended
2 LOUD specific changes, including a workaround for the lack of an amphibious category for engineeres and ACUs

Acknowledgements

Hypoxic – Highlighting an error with fort Clarke
Vortex – Replay vs M28 that highlighted a few issues with M28 (overflowing resources, overbuilding land scouts, and some other smaller points)
RD - Replay

maudlin27

v139 Update
Introducing AI variants (something I've wanted to do from back when I was working on M27 but which kept being put off for more immediate/easier changes):

M28 Adaptive - Standard M28AI - will do what it thinks is best for the situation
M28 Air - Focuses mostly on air units with minimal land units
M28 Land - Focuses mostly on land units with minimal air units
M28 Rush - Similar to M28 land but tries harder to win without getting higher tech factories
M28 Tech - tries to get away with fewer factories and more aggressive tech upgrades
M28 Turtle - Relies far more on PD and T2 arti to stop enemy attacks in the T1-early T3 stage, although will still build a modest land and air force
4 other small changes, mostly relating to trying to make M28 build a greater variety of experimentals once it gets a paragon (in case the opponent can keep up with its GC spam without suffering much damage)

I expect the AI variants will need various tweaks over time so let me know if they do anything that seems odd given their described focus. They'll also still use units the same as M28AI, i.e. the main change with the variants is to what they build, not how they use the units they build.