air strategy AI - stranges

Matrix Games Forums

Forums Register Login Photo Gallery Member List Search Calendars FAQ

My Profile Inbox Address Book My Subscription My Forums Log Out

air strategy AI - stranges

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.] >> [Modern] >> Larry Bond's Harpoon - Commander's Edition >> air strategy AI - stranges

Page: [1]

Message

<< Older Topic Newer Topic >>

air strategy AI - stranges - 12/19/2007 3:46:08 PM

project640

Posts: 5
Joined: 12/19/2007
Status: offline

Dear Matrix-Team,

Firstly,thank you for this wonderfull game. I play with this game since 10 years (from HC'97). I like because this is the only game with modern naval warfare which is so detailed. I purchased HCE edition also and I have noticed some news , like ECM (lightning), redirection of aircraft needed a time (depends from a type), soneboy usage not 100%, etc.
But, I have noticed some stranges (I want to share it , maybe will be repair in a patch

)
I played with russian side (it is more sympatic for me, because I am from "east side"

). The other opponents was two carrier battle groups, each with two carriers. After a successful heavy aircombat (versus F-14s - hardest enemy) and defense versus massive tomahawk missiles attack I have prepared a counter attack with my Tu-22 backfire bombers. Under this preparation I observed many small aircrafts in direction towards my carrier group. I launched all my wings - too late, because enemy started firing with massive harpoon attacks. After I hold this attack enemy wings -without any remaining bombs, rockets- continued their raids directly towards my DDGs. The airdefense shot they down one by one. All 36 A-6s, and S-3s!!! Why they continue attack without any arms, why they flew in sight of my DDGs! It is very easy to do if they ended their standoff attack go back to carrier. Hawkeye observed from many miles my figters despite of AI send many strikers -without any air-cover- in definite death.
With russian sides very easy after I shot down all F-14s (F-18s easy shot down with Mig-31s) I just begin attack with my backfires.
I hoped in a smarter AI, but it seems I should to wait till online game. And I am very sad because many things can repair with some works.

Post #: 1

RE: air strategy AI - stranges - 12/19/2007 4:41:01 PM

TonyE

Posts: 1551
Joined: 5/23/2006
From: MN, USA
Status: offline

Thanks project640, this is a known problem and it will be improved, but by no means perfect, in the first patch.

I'm glad you are having fun with the game

_____________________________

Sincerely,
Tony Eischens
Harpoon (HC, HCE, HUCE, Classic) programmer
HarpGamer.com Co-Owner

(in reply to project640)

Post #: 2

RE: air strategy AI - stranges - 12/19/2007 9:31:04 PM

CV32

Posts: 1046
Joined: 5/15/2006
From: The Rock, Canada
Status: offline

Its a tough job trying to get the proper balance of aggression in the AI. Right now its a crazed blood thirst, shooting all missiles and guns, and then making a suicidal rush in to poke you with a pointy stick kind of aggression. We're working on it.

_____________________________

Brad Leyte
HC3 development group member for HCE
Author of HCDB official database for HCE
Harpgamer.com Co-Owner

(in reply to TonyE)

Post #: 3

RE: air strategy AI - stranges - 12/20/2007 9:28:10 AM

project640

Posts: 5
Joined: 12/19/2007
Status: offline

Othewise I have noticed some nice events. After a massive tomahawk missiles attack I sended many ASW patrols towards missile launched zones. Between heavy air-combat my Tu-142s started lay soneboys. When AI noticed this action began attack my ASW patrols. I have got caughted two submarines (a Los Angeles and a Strugeon), but located area was too big therefore I sended more ASW in order to clairify position of submarines. Every ASWs were shot down by F-14 and both submarine got away. Just one hit with one torpedo (but it was not enough) :-(

(in reply to CV32)

Post #: 4

RE: air strategy AI - stranges - 12/20/2007 7:09:17 PM

Owl

Posts: 179
Joined: 8/4/2000
From: Portland, OR
Status: offline

I have noticed that the AI seems to take great glee in shooting up ASW helicpters and planes. It may be cheating bit, but far be it for me to take advantage of this and wait untilt the Tomcats are heading home to chase them down and kill them with any fighters I have around AFTER they've shot their missiles off.
I don't have access to the data base - are the hit % numbers for the phoenix missiles different between battelsets or do they get better over time? Some earlier scenarios seemt to have less accurate ones.

_____________________________

(.) (.)

...V...

(in reply to project640)

Post #: 5

RE: air strategy AI - stranges - 12/20/2007 7:20:49 PM

CV32

Posts: 1046
Joined: 5/15/2006
From: The Rock, Canada
Status: offline

quote:

ORIGINAL: Owl]
I don't have access to the data base - are the hit % numbers for the phoenix missiles different between battelsets or do they get better over time? Some earlier scenarios seemt to have less accurate ones.

The data for the AIM-54 Phoenix found in the HCDB database (used for the EC2003 and Westpac Battlesets) is considerably different than what you will find in the other Battlesets. The biggest difference is missile range - the old Battlesets used the full 110 nm figure, whereas this is chopped nearly in half in HCDB.

_____________________________

Brad Leyte
HC3 development group member for HCE
Author of HCDB official database for HCE
Harpgamer.com Co-Owner

(in reply to Owl)

Post #: 6

RE: air strategy AI - stranges - 12/20/2007 9:36:19 PM

SireChaos

Posts: 710
Joined: 8/14/2006
From: Frankfurt, Germany
Status: offline

quote:

ORIGINAL: CV32

quote:

ORIGINAL: Owl]
I don't have access to the data base - are the hit % numbers for the phoenix missiles different between battelsets or do they get better over time? Some earlier scenarios seemt to have less accurate ones.

The data for the AIM-54 Phoenix found in the HCDB database (used for the EC2003 and Westpac Battlesets) is considerably different than what you will find in the other Battlesets. The biggest difference is missile range - the old Battlesets used the full 110 nm figure, whereas this is chopped nearly in half in HCDB.

I noticed that to-hit probabilities have been reduced. In WinHarp they were 55% for the AIM-54 and 75% for AIM-9 and AIM-7 (IIRC) - now they are maybe half that.

(in reply to CV32)

Post #: 7

RE: air strategy AI - stranges - 12/20/2007 9:49:23 PM

CV32

Posts: 1046
Joined: 5/15/2006
From: The Rock, Canada
Status: offline

quote:

ORIGINAL: SireChaos
I noticed that to-hit probabilities have been reduced. In WinHarp they were 55% for the AIM-54 and 75% for AIM-9 and AIM-7 (IIRC) - now they are maybe half that.

That's true. The hit probabilities (pH) have been adjusted to comply with Harpoon 4 paper rules data. In most cases, that meant a reduction in pH. (Mind you, AIM-7 always had crappy pH.

)

< Message edited by CV32 -- 12/20/2007 9:50:30 PM >

_____________________________

Brad Leyte
HC3 development group member for HCE
Author of HCDB official database for HCE
Harpgamer.com Co-Owner

(in reply to SireChaos)

Post #: 8

RE: air strategy AI - stranges - 12/21/2007 4:40:12 AM

Vincenzo_Beretta

Posts: 440
Joined: 3/13/2001
From: Milan, Italy
Status: offline

I agree with the problems showed by the AC's tactical AI. I had a flight of six Norwegean F-16 execute a beautiful strike on my SAG in the GIUK 4.0 scenario "Convoy: Medium Opposed NATO Transit" (four ships sunk), only to have the same six planes merrily overfly my inner AAW defense area and being shot down one by one by the surviving ships. The only sane, and simple, tactic would have been for the planes to turn around after their missiles hit, and RTB. This changed a beautiful attack by the AI into an hard hit to my surface forces to be followed by suicide behaviour by the enemy pilots - quite anticlimatic

Since we are at it: in the same scenario I tried to detach a flight of MiG-23 flying AAW from the Banak Airbase's in a nearby sector (via the "Orders ---> Split Group" menu), but, even after detaching then from the formation, they never shoved up on the group map, and I was unable to control them . Is this some sort of bug or I'm missing something? I'm using the "Auto Setup Formations" options in the main menu.

_____________________________

(in reply to CV32)

Post #: 9

RE: air strategy AI - stranges - 12/21/2007 6:33:18 AM

hmssirius

Posts: 31
Joined: 10/15/2007
Status: offline

I think this is a good thing. The AIM 7 has a well documented history of under performance. particularly in the vietnam era. So the reduced PH may well reflect an accurate representation.

(in reply to CV32)

Post #: 10

RE: air strategy AI - stranges - 12/21/2007 11:01:51 AM

project640

Posts: 5
Joined: 12/19/2007
Status: offline

As I know the performance of these air-air missiles more or less well simulated. Launch a phoenix far than horizont is reduced the accuracy this missile. I belive it. Many times when I patrol with Mig-25R F-14s attacked with phoenix I just increased an altitude (vhigh) and speed. Ususally with this action I could avoid missiles. In israel-arab wars israel many times want to catch Mig-25s without any success. From the toher sides soviets air-aur weapons more worse then NATO oppenents it is because technical inferior (it is also believable). Hit with AA-10 versus F-18, F-14 less then 20%. Sparrow is more accurate. I was a soldier at hungarian army and hungarian pilots in every year take a missile's practice over baltic-see with aa-8 aphid. The accuracy of these missiles agains baloon was more than 90%, but agains aircraft (without any countermeasures) less than 40%.

(in reply to hmssirius)

Post #: 11

RE: air strategy AI - stranges - 12/22/2007 12:06:15 AM

Shark7

Posts: 7937
Joined: 7/24/2007
From: The Big Nowhere
Status: offline

The real problem is balancing the AI to be agressive enough to attack when it should, yet not do completely stupid things when it shouldn't.

TBH, unless you were flying directly at their base, it really is suicidal for a MiG-21 to challenge an F-15. However, this has actually happened historically, and usually with the expected results. The key is finding a way to program that type of behavior into an AI that really only sees the 2 choices...attack or don't attack. It will do one or the other every time, as opposed to attacking sometimes when it shouldn't and not attacking sometimes that it should.

Some type of aggression check would be needed, or even better, the AI agressiveness of each squadron set by the scenario designer...something we do not have control of. You could also do a complicated series of checks to determine an attack or run outcome, but at the expense of processor use. And you would still need a basic ROE for the aircraft AI by type and role.

This is something that the team wants to improve, and are trying to find ways to make it better. But finding the balance between agressive and passive AI is always a challenge.

Just my 2 cents on the matter.

_____________________________

Distant Worlds Fan

'When in doubt...attack!'

(in reply to project640)

Post #: 12

RE: air strategy AI - stranges - 12/23/2007 7:59:40 PM

SteveF2006

Posts: 111
Joined: 11/5/2006
Status: offline

I have not noticed any issue with splitting off aircraft in the way you have mentioned. I do notice that if I split off two groups in my formation at the same time, the computer shows them as one group. If this issue keeps happening, consider adding it to the "bug list" on this form!

(in reply to Vincenzo_Beretta)

Post #: 13