Matrix Games Forums

Forums  Register  Login  Photo Gallery  Member List  Search  Calendars  FAQ 

My Profile  Inbox  Address Book  My Subscription  My Forums  Log Out

Musings on moving land units

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

Logged in as: Guest
Users viewing this topic: none
  Printable Version
All Forums >> [New Releases from Matrix Games] >> World in Flames >> AI Opponent Discussion >> Musings on moving land units Page: [1]
Login
Message << Older Topic   Newer Topic >>
Musings on moving land units - 8/23/2021 3:58:25 PM   
rsunley

 

Posts: 33
Joined: 11/3/2009
From: Perth, Western Australia
Status: offline
Hi folks

For some time I have given thought to how the AIO might play the game. I have played many PC wargames over the years and inevitably one works out the patterns that is used by computer opponents, usually leading to a victorious result.

I feel that for an AIO to be truly challenging, it must try to 'think' like a human player, avoiding random actions unless a decision is truly a 50/50. To that end I have tried to analyse my own playing principles and come up with some rules that I use when gaming, whether in a PC or in the more traditional tabletop format. For a really advanced experience, an AIO would be able to analyse and learn from its play - a separate discussion perhaps.

For this post I have looked at moving land units, attempting to break down how I decide to move a unit at a tactical level. I have not taken into account the process of selecting strategic objectives, at this level I make the assumption that they have already been identified.

The tricky part is working out combinations of what both my units and the enemy units *could* do, if moved in a certain way. It is relatively easy to visualise this in my mind but for an AIO it must somehow construct a separate map where potential moves can be played out. It must then be able to get to a point where either a move is not a good idea because it violates basic principles and other units can't be moved to alleviate those breaches (for example restoring a unit to be in supply), or it is possible to satisfy the basic principles and move towards or attack its objective.

Anyway this is a first draft, I continue to think about this topic and may post more. Of course Steve and his team are likely to have their own ideas which may be far better and easier to code than these idle notes.


--

Considerations about moving land units

A unit must always have an objective hex
If it is not occupying its objective hex, it moves towards it by the route that uses the least movement points
If a unit cannot reach its objective due to enemy units, assess them to be attacked
Unsupplied units may move but are subject to the counterattack principle as per other units.

Each hex has a defensive ratio defined by terrain (DR)
The DR is calculated by working out the averaged combat effect (CE) of every surrounding attackable hexside (e.g not high mountain if no mtn can reach) that can potentially be reached or is occupied by opposing units as a ratio against the DV of the defending hex.
Hex DV values
Clear, Desert, Ice, Tundra 1
Mountain, Swamp, Desert Mountain 2
Forest 1.25
Jungle* 1.5 (1.25 against Japanese units)

Hexside CE values
No features, borders, weather boundaries 1
River , Canal, Regular Fort, Straits*, High Mountain*, Lake*, All Sea* 0.5
Improved Fort 0.33

*Denotes only applies to certain units, or only permitted for some units

The maximum DR for a hex is 1 (clear/desert/ice/tundra surrounded by 6 similar hexes). Mountains surrounded by rivers would have a much smaller ratio for example. The values for Forest and Jungle attempt to take into account the effect on air power and Japanese jungle ability. Perhaps some tweaking could be done for terrain that allows the defender to choose the assault table.

Overall principles
Always use all of a units movement points unless enemy ZOCs prevent further advance or target is reached
Do not move a unit to an unsupplied position unless another friendly unit can be moved in the same impulse to restore supply
Do not move a unit to a hex where it can be counterattacked at 4:1 (A) or 3:1 (B) unless another unit can be moved in the same impulse to the same hex or the hex is a strategic objective (city, resource, factory)
Do not move a unit to a hex where it can be placed out of supply by enemy moves unless the hex contains a strategic objective or another unit can be moved to prevent enemy units from cutting supply
Select a target hex with a higher defensive ratio in preference to a lower one if it is unoccupied (for example, a mountain hex), unless the target contains a strategic objective

Defensive principles
Cities, factories and resources are always to be defended, never abandoned
Choose hexes with the best possible defensive ratio and adjacent to other friendly units (existing or potential) where possible
Major ports are always defended if the enemy can potentially move naval units to an adjacent sea area
Minor ports in a units home country are always defended if the enemy can potentially move naval transport units to an adjacent sea area
Germany defends non-port coastal hexes adjacent to the North Sea. The number of units depends on the number of land units in Britain
The CW defends non-port coastal hexes in Britain if there are German land units stacked with transports that can reach the North Sea

Attacking principles
4:1 (A) / 3:1 (B) is the general threshold for a land attack (the AIO always attacks at these or better, subject to action limits).
If odds of potential attacks are even, select single units ahead of stacked ones and disorganised units ahead of organised. Units occupying a strategic objective rank higher than those elsewhere. A Blitz attack that can eliminate the defender with a retreat is assessed at 1 odds level better than the raw values.
Attack with as many units as can reach an attacking hex. If there are units that can reach multiple combats, calculate if an increase in odds level can be obtained by committing any or all such units.
If the potential defensive Tactical Bombing Factors that could be delivered to the hex bring the odds below the general thresholds, do not attack. Do not consider any interception but do calculate for terrain and weather.
Add the values of attacking TBFs (modifying for terrain and weather) that can reach the target without being intercepted to the attackers total factors
Add the values of any available shore bombardment factors (including defensive if option has been chosen)


Decision making
A chain of moves may be required in order to meet the principles regarding supply and counterattack. If it is not possible for every unit in the chain to make a satisfactory move (i.e. not violating any principles) then all moves in the chain are undone and the target action must be reassessed.

When calculating possible enemy moves, record against each reachable hex the details of every enemy unit that could enter it. Recalculate this after a potential move is evaluated, taking into account changes in ZOCs. Do the same for friendly moves, maintaining a list of possible friendly units that could reach a hex. If a potential move is made, update the surrounding hex information accordingly.
In effect this means having a pseudo-map for potential moves that is constantly updated and analysed.


Headquarters
Do not attack unless doing so brings an attack to 7:1 (B) or 10:1 (A) or it is impulse 9 or greater
Move so that the greatest number of unsupplied units are placed back in supply, if any
Move so that the greatest number of unsupplied hexes that can be potentially be reached by friendly units are placed into supply
Never move to a hex that is out of supply, unless it cannot reach any hex that is in supply. If it is currently supplied it moves to the hex with either a strategic objective or the best defensive ratio and stacks with another unit if possible. If it is currently unsupplied it does not move.


--

Ralph Sunley
Post #: 1
RE: Musings on moving land units - 1/14/2022 8:25:23 PM   
Neilster


Posts: 2890
Joined: 10/27/2003
From: Hobart, Tasmania, Australia
Status: offline
The AIO analysing its own play is a massive task and would possibly involve neural networks. The AIO will rely on the distilled experience of WiF experts.

It sounds like you're proposing a system where strategic objectives are decided on and then there is a "bottom-up" approach to achieving them. Steve has already decided that a bottom-up approach leads to a combinatorial explosion of potential moves and is hence infeasible. My understanding is that more of a "top-down" approach will be pursued, although the sort of considerations you have outlined are still relevant at the tactical level.


< Message edited by Neilster -- 1/14/2022 8:33:04 PM >


_____________________________

Cheers, Neilster

(in reply to rsunley)
Post #: 2
Page:   [1]
All Forums >> [New Releases from Matrix Games] >> World in Flames >> AI Opponent Discussion >> Musings on moving land units Page: [1]
Jump to:





New Messages No New Messages
Hot Topic w/ New Messages Hot Topic w/o New Messages
Locked w/ New Messages Locked w/o New Messages
 Post New Thread
 Reply to Message
 Post New Poll
 Submit Vote
 Delete My Own Post
 Delete My Own Thread
 Rate Posts


Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI

0.641