Matrix Games Forums

Harry Erwin, PhD.......Computational neuroethologist:

ORIGINAL: herwin

ORIGINAL: floydg

ORIGINAL: herwin

I like Bohdi's utility in large part because it organises intel data similarly to what I'm used to. I'd like to see an effective moving ships tracker algorithm added, but that would be a really big deal, involving input from my current research in autonomous robot control.

I'm not sure what you mean by "moving ships tracker algorithm". Do you mean tracking the movement of enemy ships?

The knee-jerk engineering solution would be a Kalman filter, but that has a series of weaknesses:
1. A Kalman filter is an optimal linear filter, but ships follow curved paths (geodesics or rhumb lines).
2. Ships maneuver or sail to waypoints and change course. Kalman filters are too dumb to pick up on situational clues or take account of how ship captains actually maneuver their ships.
3. The expansion of the covariance matrix for the Kalman filter over time does not model how ship tracks become uncertain after a few days off the screen.

On the other hand, a non-linear filter can be faked out by deceptive tactics and will be sensitive to the order and timing of updates to tracks. Association of track reports to the correct track is also a hard problem--we run into this in auditory scene analysis.

So I use the Bohdi utility and draw my own conclusions from the pattern of reports. If you provide something similar, the reports should be colour-coded in some obvious way based on how old they are.

ORIGINAL: Mynok


I totally glazed over about word 5.........then read this, explaining all:

Harry Erwin, PhD.......Computational neuroethologist:

[:D]

ORIGINAL: floydg

I doubt we'll be able to solve that problem for you with this program. I'm sure if we could reliably predict the path of a task force, nobody would want to play against us (or we'd be doing some super-secret government work we can't tell you about).

Floyd

ORIGINAL: Reg

ORIGINAL: floydg

I doubt we'll be able to solve that problem for you with this program. I'm sure if we could reliably predict the path of a task force, nobody would want to play against us (or we'd be doing some super-secret government work we can't tell you about).

Floyd

I don't know about the rest of you guys, but until the game can produce task force tracks like the diagram below, I am not sure that I would want a tool which will provide intel to the players which will be much more accurate than what would have been available to contemporary commanders.

(Not to mention play balance issues against players who are not using the tool if it is not included as part of the standard distribution).

I am aware that AE is introducing way-points which will partially address the issue but I suspect it will not go far enough to disguise paths at the scale of this game if this tool is too effective.

ORIGINAL: floydg

Are you asking for a history of where a TF was or a prediction of where it will be? I think Harry was asking for the latter, but we could certainly do the former (and I believe witpDecoder already does that).

Floyd

ORIGINAL: floydg

ORIGINAL: Reg

ORIGINAL: floydg

I doubt we'll be able to solve that problem for you with this program. I'm sure if we could reliably predict the path of a task force, nobody would want to play against us (or we'd be doing some super-secret government work we can't tell you about).

Floyd

I don't know about the rest of you guys, but until the game can produce task force tracks like the diagram below, I am not sure that I would want a tool which will provide intel to the players which will be much more accurate than what would have been available to contemporary commanders.

(Not to mention play balance issues against players who are not using the tool if it is not included as part of the standard distribution).

I am aware that AE is introducing way-points which will partially address the issue but I suspect it will not go far enough to disguise paths at the scale of this game if this tool is too effective.

Are you asking for a history of where a TF was or a prediction of where it will be? I think Harry was asking for the latter, but we could certainly do the former (and I believe witpDecoder already does that).

Floyd

ORIGINAL: Roger Neilson II

Couple of suggestions

I hate the info screens as you can't take info from them unless you capture them as screen shots. In particular I'd like to be able to get the ships sunk data and use it in other ways.... I'm not after any info I should not have, but it would be good to take info like this and be able to put it into AARs.... with comments. At present you can't select a list, but have to settle for whatever is shown on the screen shot and this limits what you can do with it. It would also allow greater annotation and dating of such info if extra text fields were available. For example the four capital ships here - three were lost in the initial week of the war..... only the Pennsylvania was lost later in defence of Menando.

Roger

ORIGINAL: Coach Z

Looks very interesting