Sims for Training

[Erwin]

Interesting discussion excerpts

"For a fully automated sim tool the AI has to be smart enough to do a reasonable job of managing subordinates. The behaviours have to be realistic. The units have to be situationally aware and able to:

• develop realistic plans based on an assigned objective, including the ability to:

o assess the situation (terrain, enemy, friendlies)

o develop courses of action (COA) based on doctrine

o allocate resources across tasks

o schedule tasks

o analyse and select the best COA

• issue orders to their subordinates

• reschedule tasks if progress changes (both slip and crib)

• react to developments as they occur (eg take cover, return fire, retreat if need be, call in fire support, advise their boss if they are in trouble etc),

• periodically reassess their plans (eg plot a new bypass route if enemy too strong ahead, bunker down if under serious threat, switch from Move to attack if so authorised, pull back to a secondary blocking position if delaying etc) and if need be re-send orders.

In short they have to be able to implement a realistic military decision process and cycle. The base mechanics of fire, move, rest and reorg have to be modelled well. Modelling of soft factors like morale, cohesion and fatigue are just as essential as hard factors like weapon effectiveness.

I’m not saying it is easy. It’s not and it takes a lot of time. But it is do-able. We have spent over fifty man years of effort getting our Command Ops engine to the point where it does this and we still get a lot of feedback from users that this needs tweaking or that is not quite right. But overall our user base which comprises a very large proportion of military and ex-military users are happy with the AI’s performance especially in regard to its management of subordinates. This frees them up to focus on what commanders should be focussing on and not the minutiae of controlling every sub unit."

"IMO the sim community within Defence should be focussing on those areas that

> are going to make compelling arguments for funding. Focussing on COIN when

> the political will is moving away from it won't make for a compelling

> argument. That is not to say that at some point in the future it won't be

> needed but it's not going to garner precious funding under the current

> climate.

>

> To my mind there are two compelling argument for funding the use of

> simulation within the military. First and foremost is the provision of a new

> warfighting capability to support operational decision making. More on that

> in a minute. The second one is the cost savings derived from using sims in

> training and experimentation & analysis.

>

> The cost argument appeals to the managers and accountants but it's the

> decision support argument that will appeal to the new breed of operational

> war-fighters. This is the sexy argument and the one that has the greatest

> chance of going forward. There are quite a few older commanders who are not

> comfortable with the use of simulations in the operational decision support

> role but certainly the younger commanders and many of the senior ones too

> that are willing to embrace it. The Chief of Defence Force in Australia is

> an avowed advocate of the use of simulation.

>

> Both of these roles require a low overhead operational warfare engine. The

> days of the high overhead SAF are numbered. They cost too much to run and

> are too slow to be effective as on operational decision support tool ( other

> than in pre op planning, where there is sufficient time and resources ). To

> be effective in the operational decision support role you need a sim tool

> that can take a commanders intent, prepare three courses of action (COA),

> wargame them, analyse them and prepare a report with recordings to

> demonstrate the results and the recommended COA for the commander's review.

> Once the commander has made their decision and modified the plan as they see

> fit, it needs to be able to upload the plan to the C2 command boxes for

> implementation. All this needs to be done in two hours at a Brigade HQ. SAF

> products cannot meet these requirements. Only a fully automated low overhead

> sim tool can."