Observations on AI Behaviour / Bug or Design?

Matrix Games Forums

Forums Register Login Photo Gallery Member List Search Calendars FAQ

My Profile Inbox Address Book My Subscription My Forums Log Out

Observations on AI Behaviour / Bug or Design?

View related threads: (in this forum | in all forums)

Logged in as: Guest

Users viewing this topic: none

Printable Version

All Forums >> [Current Games From Matrix.] >> [World War II] >> War In The Pacific - Struggle Against Japan 1941 - 1945 >> Tech Support >> Observations on AI Behaviour / Bug or Design?

Page: [1]

Message

<< Older Topic Newer Topic >>

Observations on AI Behaviour / Bug or Design? - 8/27/2004 11:53:19 PM

Rainer

Posts: 1210
Joined: 11/21/2000
From: Neuching, Bavaria, Germany
Status: offline

Scenario 15 (The BIG one), Japanese Computer, latest version (all patches applied), SE Human, all other areas Computer, Date July 8, 1942, Daily Turns

I want to share some observations of AI Behaviour to help find out what's "buggy" and what has been implemented intentionally "by design". One or the other design decision may or may not need reconsideration.

1. Auto Routes
Most, but not all convoys, either Auto-Convoy or convoys set up manually, avoid the (mid 1942) dangerous area of the Marshall Islands (Kwajalein etc.). About 2 out of 10 convoys stray too far to the West, are spotted, attacked and sunk by Japanese aircraft (happened twice, then I interferred and routed those convoys manually around the danger zone).
This happens nearly always on the way from the US coast to the Allied SoPac/SWPac bases. Returning most of these TFs choose a route much more to the South, thus no danger.

2. Med Bombers and Recon
It seems not to be possible to make an air unit attack a specific target.
Example: Trying to plan an attack out of Rabaul towards Gili Gili I ordered Moresby based Catalinas to recon Gili Gili, even limited deliberately their range. Still the Cats either went elsewhere or (with the appropriate range limit set) did not take off at all.
Same with Moresby based medium bombers (Marauders and Bostons). Ordered to ground attack Gili Gili they simply ignore those orders and attack Lae or Buna.
Even after landing troops at Gili Gili and starting the fight the bombers don't care and ignore specific orders to help in the struggle. This is rather strange, and not very realistic. From bombarding Gili Gili I knew that fairly strong Japanese forces were there. But even if those enemy forces were weak or not existent (how does AI know about this without cheating?) at least the ordered recon missions should be carried out, lest the commander (player) is kept in the dark about what he/she is running into.

3. Planning (Future Objective)
Setting "Future Objective" apparently does not work. In order to attack Gili Gili (the example above) 6th Aus Div and 30th Aus Brig were set to plan for Gili Gili. Every 3rd turn or so I found that the objectives had been reset to Moresby, starting over at zero.

4. Straying Ground Units
I have seen at least one unit (80th Engineer Aviation Bataillon) starting to move on its own towards Gili Gili. I did not order this and it is totally unclear why this unit decided to march to Japanese occupied Gili Gili (not to mention the cumbersome route across the mountains). When I saw them doing this I immediately ordered them back to Moresby. Next turn they again turned to Gili Gili (on their own), and I left them this time do what they "want" because Gili Gili had been conquered/liberated.

Please note: ALL units mentioned above are under FULL HUMAN control, as is the area (SE). And yes, I double checked the units in questions are under "Human Control".

Any comments/advice welcommed.
Rainer

Post #: 1

RE: Observations on AI Behaviour / Bug or Design? - 8/28/2004 3:05:19 AM

DoomedMantis

Posts: 1922
Joined: 8/24/2002
From: Sydney, Australia
Status: offline

The AI is being looked at by the developers, specifically regarding your point 1 (but other things as well). As for your other points, do you have a save that repeats the problem from a save from before it does it? If so send it onto me and I will have a look as they shouldnt be doing that.

paul.harrison50@bigpond.com.au

_____________________________

I shall make it a felony to drink small beer.

- Shakespeare

(in reply to Rainer)

Post #: 2

RE: Observations on AI Behaviour / Bug or Design? - 8/28/2004 9:10:16 PM

Rainer

Posts: 1210
Joined: 11/21/2000
From: Neuching, Bavaria, Germany
Status: offline

Last three saves via email on the way to you around the globe :)
Cheers
Rainer

(in reply to DoomedMantis)

Post #: 3

RE: Observations on AI Behaviour / Bug or Design? - 8/29/2004 9:39:23 AM

DoomedMantis

Posts: 1922
Joined: 8/24/2002
From: Sydney, Australia
Status: offline

got them

_____________________________

I shall make it a felony to drink small beer.

- Shakespeare

(in reply to Rainer)

Post #: 4

RE: Observations on AI Behaviour / Bug or Design? - 9/11/2004 8:14:39 AM

SeaJunk

Posts: 30
Joined: 7/21/2004
From: CA, USA
Status: offline

I'd like to second the sw pacific LCU weirdness, V 1.21 scen 14.

I also have the 6th aus div in port moresby that I directed to buna. After moving one hex it resets its target and DH back to moresby. If I try to send it back to Buna, it resets to moresby during turn resolution.

The 52 base force also marched by itself from moresby to gili gili. Again no amount of trying to send it back to moresby works. Gilli is an allied base owned by sw pacific, 10/42. Zones and bases under human control.

Looking forward to 1.3.

Thanks
SJ

(in reply to DoomedMantis)

Post #: 5

RE: Observations on AI Behaviour / Bug or Design? - 9/11/2004 10:26:47 PM

Buck Beach

Posts: 1973
Joined: 6/25/2000
From: Upland,CA,USA
Status: offline

Major AI or bug disappointment here! Playing Lemur's mod under 1.21 as the Allies. I'm at mid March 1942 and I am (at least was) enjoying the hell out of all the action in the DEI and Burma and the building up in the SWPac and SoPAC while battling tooth and nail with the Jap subs of the coast of OZ. I have 2 groups of bombers, 3 P-40Es squadrons and a Catalina group in PM. Now I gave some thought to the lack of the Jap response to all my (constant activity going in and out of PM building supplies up and milk run bombings of Jap bases and shipping, but focused my attention to the other areas. Finally, I decide to expand Gili Gili with another small base unit that was occupied by the alllied forces out of Rabual since the first couple of week of the war. Sent in an AP and an AK in separate TFs. Curiousity got the best of me and I switch sides to see "What the F---" was going on.

Rabaul (only base in the area) is at level 4, has 55,000 plus in supplies, 2 groups of high experinced Betties set on Naval Attack and rest, 3 groups of high experienced Zeros for a total of 45 plus AC (all set on LR Cap) and 2 groups of recon/float planes. The Allied TFs at Gili are spotted.

There has been NO attacks out from them the whole game thus far, albeit they have reconed Buna on numerous occasions. This is a game breaker for me. I know eventually the game will be fixed by Matrix but I am stopped in my tracks for now and am very frustrated. I'm not opposed to starting the game over but for what?

BTW, I re-ran several other previous turns again only to have similiar results or lack there of.

< Message edited by Buck Beach -- 9/12/2004 1:03:51 AM >

(in reply to Rainer)

Post #: 6

RE: Observations on AI Behaviour / Bug or Design? - 9/12/2004 3:35:39 AM

Twotribes

Posts: 6929
Joined: 2/15/2002
From: Jacksonville NC
Status: offline

Rabual regularly attacks Port Morseby in my game, and has sortied against my Carriers when I made the mistake of moving into their range. They have also sunk AK and damaged TK when in range.

(in reply to Buck Beach)

Post #: 7

RE: Observations on AI Behaviour / Bug or Design? - 9/12/2004 4:19:52 AM

Buck Beach

Posts: 1973
Joined: 6/25/2000
From: Upland,CA,USA
Status: offline

quote:

ORIGINAL: Twotribes

Rabual regularly attacks Port Morseby in my game, and has sortied against my Carriers when I made the mistake of moving into their range. They have also sunk AK and damaged TK when in range.

What scenario are you playing and what is the game date. Are you playing Lemur's mod of 15?

(in reply to Twotribes)

Post #: 8

RE: Observations on AI Behaviour / Bug or Design? - 9/12/2004 4:38:45 AM

Twotribes

Posts: 6929
Joined: 2/15/2002
From: Jacksonville NC
Status: offline

I am play scenario 15, and no not Lemur's.

(in reply to Buck Beach)

Post #: 9

Page: [1]

All Forums >> [Current Games From Matrix.] >> [World War II] >> War In The Pacific - Struggle Against Japan 1941 - 1945 >> Tech Support >> Observations on AI Behaviour / Bug or Design?

Page: [1]