I'd like suggestions on modelling fields of small, unmanned weapons platforms
that would move slowly towards intruders, with the idea of overwhelming by
numbers.
I'd need them to track targets, build to killing mass against anything that
would stick around long enough, but relatively ineffective against
hit-and-run.
Please note that I'm most interested in suggestions on modelling
'behavior', auto-rules for a relatively amorphous pattern of slow-moving
platforms.
The_Beast
Model it as a large amorphous organic being - the field is a certain
size and density and the whole field moves to try to place the target into the
center of the being. Once the target is in the center, it is attacked by mines
until destroyed.
Give the field a sensor range and a move rate, then a number of attacks based
on density.
You can then vary minefields by size, density, move rate, attack rate and
sensor capability.
--Binhan
> -----Original Message-----
> devans@nebraska.edu wrote:
> I'd like suggestions on modelling fields of small, unmanned weapons
This sounds similar to an AI behaviour known as 'flocking' (see
http://www.red3d.com/cwr/boids/ if you're interested in details). This
is fairly straightforward to implement when you have a powerful CPU updating
fifty times a second, less trivial with a pocket calculator and
the granularity of an FT turn!
I'd suggest that the 'bots' (for want of a better term) don't use either
of the FT movement systems - off the top of my head, I think
auto-manoeuvering rules for either could be nightmarish to work out
(although I could be wrong).
I'd suggest defining one bot as a 'swarm leader', and he moves by a set
of rules - which could be as simple as "always move towards the target".
The others bots in the swarm would merely keep formation with the leader
instead of having their own rules (a bit like moving the squad leader in
SG2 and placing the rest of the squad around him). Speed of the swarm would be
low and I guess they'd effectively move like fighters. Alternatively, each
swarm member could obey a simple rule like "always move towards the swarm
leader".
If they're armed with (for example) a turreted beam weapon with 360°
field of fire, their facing won't matter, if they have a facing-critical
weapon then you could assign them a fixed turn rate (say 3-4 clock
points per turn) which is used with another simple set of rules ("always
turn towards the target").
If the leader is destroyed, you could have a turn or two of confusion (each
bot moves in a random direction and changes facing by a random amount) before
another swarm member assumes leader responsibilities. This could make it easy
for the opposition to target the leader and nullify the swarm so you could
employ some subterfuge (randomly reassign
the leader each turn?).
If multiple targets appear, multiple swarms would choose targets
randomly - I'm sure you could work out a system that biases towards
closer targets but stops every swarm picking the same target.
Just a few thoughts - back to programming some real game AI !
> Model it as a large amorphous organic being - the field is a certain
> Give the field a sensor range and a move rate, then a number of attacks
> You can then vary minefields by size, density, move rate, attack rate
I was thinking of individual 'mines', but I'll give this a look, maybe with
the field broken into sectors, each being an organelle. Might give me the
effect of concentrating if you aren't scooting through the battle area.
The_Beast
The_Beast said:
> I'd like suggestions on modelling fields of small, unmanned weapons
Vector or cinematic will make a difference, but generally I'd say something
like: a. maintain a minimum 2" disperal between units, while b. moving toward
the Unmodified End Point of the target, and c. attempt to match vectors with
it
Excellent! I worried I wasn't clear enough, but you've seen through my
mumblings quite nicely!
***
This is fairly straightforward to implement when you have a powerful CPU
updating fifty times a second, less trivial with a pocket calculator and the
granularity of an FT turn!
***
As I assumed, but sometimes such complicated 'behaviours' can be approximated
with a set of fairly simple rules. I crossed my fingers and
asked...
***
I'd suggest defining one bot as a 'swarm leader', and he moves by a set
of rules - which could be as simple as "always move towards the target".
The others bots in the swarm would merely keep formation with the leader
instead of having their own rules (a bit like moving the squad leader in SG2
and placing the rest of the squad around him). Speed of the swarm would be low
and I guess they'd effectively move like fighters. Alternatively, each swarm
member could obey a simple rule like "always move towards the swarm leader".
***
At first blush, this sounds like field-as-being concept in effect. I'll
cogitate on it more, though.
And, yes, I was figuring on 360 degree fire and movement to keep things
simple, though I thought I'd still include either cinematic or vector inertia.
***
Vector or cinematic will make a difference, but generally I'd say something
like: a. maintain a minimum 2" disperal between units, while b. moving toward
the Unmodified End Point of the target, and c. attempt to match vectors with
it
***
Getting closer to what I had in mind for a description, though I'll also have
to 'cogitate' on this. At the least, as this is a particular area that's
guarded by the field, the units would have a 'call back' set of instructions.
On the other hand, this starts getting almost hopelessly complex with
multiple, dispersed bogies, and once you add levels-of-threat...!
However, the scenerio percholating in me brain requires something that will
have this complexity to have flaws exploited by particular lines of timed
attacks.
The Federation base may YET fall to the Cardassian fleet!
The_Beast
I said:
> Vector or cinematic will make a difference, but generally I'd say
And what's really worrisome is that these are the same general rules that the
guy who wrote the Boids program came up with. If an English major can come up
with it, independently, in about a minute's thought....
Da Beast said:
> have to 'cogitate' on this. At the least, as this is a particular area
"Do got go further than X from FieldCenter point"
DaBeast continued:
> On the other hand, this starts getting almost hopelessly complex with
Head towards the closest qualifying target. A target is qualified if it
does not already have 1 mine/M mass attacking it (adjust M depending on
how deadly the mines are).
Could get complicated if you have a lot of mines and treat each
individually--I'd say group move is indicated.
Instead of one large field, you can have multiple mine field entities -
some equipped to sense large ships, some for smaller ships. Smaller ships
would be attacked by more mobile, small fields and larger ships would be
attacked by larger, slower moving fields. Since each entity has a different
criteria for targets, the fields would move in different directions at
different targets. Fleets with mixed ships would attract multiple fields.
To keep the fields "tethered" in a flexible way, instead of a hard limit
before the field returns, place a flexible limit, such as the distance to the
target is less than twice the distance to the defence point, otherwise the
field stops or retreats. This would give wide ranging fields that slowly
concentrate to the defense point. It also means, the more you stir things up,
the more concentrated the field becomes, so a quick in and out dash would be
more effective than a bunch of probing
maneuvers. Or to really confuse matters, mix and match - some fields
have hard limit and others have a flexible limit.
--Binhan
> -----Original Message-----