Jump to content

Best Way to Set-Up a Completely Autonomous Scenario


DLDC_Sims_guy

Recommended Posts

SB has a very powerful set of AI tools, obviously. It is very easy to set-up a "smart" enemy for SP and CO-OP missions. I am, however, finding it more difficult to script an entire battle for the AI to run completely from start to finish. Some general questions I have:

1. What is the best way to move forces? Should each unit be given specific routes, or would it be best to use individual waypoints and use the "and/or" dialogue to move units?

2. Does the AI use of artillery (on or off-map) have any rhyme or reason to it, or does it just wait until a unit for that side spots an enemy unit (when set to either Observer units or any units spotting) and just throw everything at it? I know you can set some very general parameters, such as priority of fires. Does having on-map arty provide more flexible arty fire when run autonomously (i.e., no manual calls for fire) using the "and/or" dialogue?

3. Is there an in-depth tutorial available on setting-up triggers and such?

That's about it for now...but it seems enough for one post which could likely be several different threads.

Curt

Link to comment
Share on other sites

1. Setup up zones and other control measures to reflect the real world idea of NAIs, TAIs and DPs (Named Areas of Interest, Targeted Areas of Interest and Decision Points - no idea if those concepts are familiar to you or not) so if the enemy is observed in NAI 1 embark on route X, if the enemy is seen in NAI 2 embark on route Y.

2. Again script artillery usage in the editor. Enemy seen in NAI 1, wait X minutes (reflecting the estimated rate of movement) and fire mission into TAI 1.

3. Not so much "in depth" (of course that term is relative) but Toyguy's YouTube series would probably help you.

Link to comment
Share on other sites

That was what I was planning to do (I know all of the doctrinal terms and their functions), I just have little idea of where to start with them. The primary problem is the terrain - very wooded, few avenues of approach, and terrain hilly enough to prevent any kind of long-range spotting (most often down to 200-300 meters!).

This will definitely be a challenge!:c:

Link to comment
Share on other sites

Well I'd have thought that would make it easier. :)

Restricted terrain means limited avenues of approach mean you don't need as many NAIs.

Similarly you need few dismounts to provide "eyes on" those NAIs and to trigger responses.

You can even use "events" to generate contact reports from things like unattended ground sensors or ground surveillance radars and generating these events can also start ground forces or UAVs moving (embark if ... event X is true, etc.) or fire missions into TAIs.

Link to comment
Share on other sites

  • Members

Some ideas.

The more fluid the battle is conducted by each party, the more difficult it becomes to script. If in doubt, go with robust (=simple) decisions. Rather than checking if a certain unit has reached a specific waypoint, check whether (own) forces in a minimum strength of X have occupied a certain region for a minimum time period (say, one minute or more).

If a human player is supposed to make the "big" decisions, Triggers are a good choice. I call triggers "the solution for problems that require eight arms and a human brain" - coordinating simultaneous actions that will start upon a human player's will (when, and what). Since the activation of a (blue) Trigger can also be referenced in a (blue) Event, and since other parties' Nameable Events and Conditions can reference blue Events, you can this way also control actions of other parties by setting blue Triggers (I don't think that you can directly reference blue Trigger states, just Events and Conditions).

Let's say that you let the entire blue task force take "the left approach" by activating Trigger 51. Blue Event 51 will become true if Trigger 51 is set. You can then fake Red's appropriate response to Blue's move by using the blue Event 51 as a reference for the red task force's own movement (maybe with a delay of ten minutes or so that they might need to figure out what their opponents are doing now). Combine this with a random element and set a less adequate response with a lesser likelihood and you could also fake a response that opens an opportunity for Blue to exploit its initial advantage.

Link to comment
Share on other sites

Well, this scenario requires absolutely NO human interaction, and that is where the difficulty comes in. It would be easy if I could assign units to actionable maneuver graphics (it's not...is it?). As it is, even some simple tests of the pathfinding AI are turning-out rather, umm, muddled...

Link to comment
Share on other sites

  • Members
Well, this scenario requires absolutely NO human interaction, and that is where the difficulty comes in.

...because you are reassigning a simulation designed for TRAINING (humans) into an analytical tool (which it isn't). I know that the other tools at your disposal aren't well suited for the task either; I just want to clarify that what you perceive as functional deficits are features that weren't implemented because they were no design requirement to begin with.

You will have to compromise, one way or the other. SB Pro gives the user a by far wider span of control than certain first person shooter oriented simulations but it isn't designed to run without human interaction at all except for rather basic and short scenarios (like the "00 Introduction" tutorials for each vehicle). You could opt to run fewer iterations and have someone to support each run by activating certain triggers and maybe manually resolving the one or other traffic jam, or limit yourself to smaller scale scenarios, or pick certain other equipment.

Link to comment
Share on other sites

...because you are reassigning a simulation designed for TRAINING (humans) into an analytical tool (which it isn't). I know that the other tools at your disposal aren't well suited for the task either; I just want to clarify that what you perceive as functional deficits are features that weren't implemented because they were no design requirement to begin with.

You will have to compromise, one way or the other. SB Pro gives the user a by far wider span of control than certain first person shooter oriented simulations but it isn't designed to run without human interaction at all except for rather basic and short scenarios (like the "00 Introduction" tutorials for each vehicle). You could opt to run fewer iterations and have someone to support each run by activating certain triggers and maybe manually resolving the one or other traffic jam, or limit yourself to smaller scale scenarios, or pick certain other equipment.

I have dealt with this kind of AI before, and it is quite possible to do exactly what I want. I have already created a small engagement with a platoon of scouts dug-in against a company of enemy scouts and a company of infantry. I have arty calls set-up to fire with certain criteria, fire controls set, etc. Right now everything is looking good, and I can get a minimum of 50% attrition, which is acceptable considering the red plan and the terrain. I have even figured out a way to replicate the 120mm laser-guided tank-killing mortar round the Swedes have.

Anyway, the scenario will eventually be much simpler than I am making it out to be. It is a delaying action, and much of the defending forces will be dying in place, rendering major displacement moves essentially useless. Amazing what just a couple of simple minefields and some felled trees can do to an attacker in restricted terrain...:wink2:

Link to comment
Share on other sites

The variable tool works wonders for randomizing AI behaviors.

For offensive missions set multiple paths, using the "embark if" and variable tool.

This will randomize your AI avenue based on the variable result.

For defensive missions set multiple paths to defensive positions and use the "move to the end" and the random variable tool.

This will warp/jump the AI directly to the defensive position at random locations.

Using the variable tool and your imagination you can literally script 100 different courses of action.

Another reason why the SB mission editor is the best thing I've seen.

Link to comment
Share on other sites

  • Members

It's important to realize that the randomness can be coordinated. The value of each random variable X1, X2, ... X63 will remain CONSTANT during the course of a scenario. So, if unit 1/A takes the left route if 0 < X3 < 50, it'll do so in 50% of all cases that you run the scenario. But you also probably want 2/A and 3/A to take a route that supports 1/A's coice of "going left", so you reference the same random variable X3 with the same quartile (from 0...50) to get coordinated random behavior. Using a different variable or a different quartile or the variable XNew will just create chaotic randomness, typically unwanted.

Link to comment
Share on other sites

  • 3 months later...

Just an update. I completed two scenarios (Swedish INF BN reinforced with a tank company delaying action vs. a Russian Mech Bde in attack, and meeting engagement between the same). I set each to run at max time limit +20% random, but most battles were resolved in under 2 hours.

These were completely autonomous scenarios run 20x each with very little input, primarily to destroy a damaged vehicle blocking movement of everything behind it. The results are being used for a Master's Thesis by a Swedish Major here in CGSC.

If anyone is interested, I will see if I can post his thesis as well as the data from which some of his paper was drawn.

Curt

Link to comment
Share on other sites

  • 3 weeks later...
Well it might be helpful to know what the Title / topic actually was?

"The impact of the European swallow early nesting on AFV manoeuvre" doesn't do that much for me, regardless of what the modelling tool is. :)

The topic is asking whether or not Swedish doctrine (a massive counterattack against an invader) is consonant with Swedish military reality (forces reduced from 32 brigades to 2 over the past 10-20 years).

The SB scenarios, that Curt put a huge amount of effort into, give the student a chance to compare outcomes from different tactical approaches.

I expect we (I'm the thesis committee chair) will be able to post the paper and the scenarios, or links to them, once it is completed - it has to be done by the middle of June. We jumped through a number of hoops to keep this thesis unclassified and unrestricted.

Link to comment
Share on other sites

The results are being used for a Master's Thesis by a Swedish Major here in CGSC.

Is this CGSC here in the US? If so, and they have SB Pro that the students work with, that is awesome. I'd been kind of bummed that I had not heard more about SB Pro use within the organizational US Army (given all the uses I could think of for its use), outside of those using SB Pro PE on their own. Maybe I've just not been reading the right threads.

Doug

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...