The Dreadnought Project Forum Index The Dreadnought Project
Naval History in the years 1890-1920
 
 FAQFAQ   SearchSearch   MemberlistMemberlist   UsergroupsUsergroups   Reg1sterReg1ster 
 ProfileProfile   Log in to check your private messagesLog in to check your private messages   Log inLog in 

Adding more...

 
Post new topic   Reply to topic    The Dreadnought Project Forum Index -> Simulation
View previous topic :: View next topic  
Author Message
tone
Site Admin


Joined: 29 Oct 2004
Posts: 479
Location: Boston

PostPosted: Fri Apr 01, 2005 1:03 pm    Post subject: Adding more... Reply with quote

I have the code on the blocks at the moment and am working to add ballistic drift and the lateral drift errors caused by wind. The Sight Manual of 1916 has given me a greater understanding of how various sights corrected for drift and the Range Tables of 1910 and 1918 offer at least a few empirical accountings for the degree to which drift affected various shells.

As a rough guide, a shell will be accelerated laterally at around .2 meters per second squared in the direction the top of the shell is spinning as it flies downrange. Therefore, the lateral displacement is proportional to the square of the time of flight (plus or minus small factors that would be difficult to characterize precisely).

A very few sights had cams within them whose shape applied the correct yaw correction as range was set -- some of these corrections may have been automatic, but in other cases it seems the index mark on the deflection dial would move to indicate a proper additional deflection adjustment that would get the drift factored out.

More commonly, the sighting assemblies upon which the telescope carriers were mounted was "inclined" or rolled 2-3 degrees to the left. In such an arrangement, an imperfect but helpful correction would take hold the more the scopes were depressed as range was dialed on. This imperfect approach was apparently the basis of the need for the "uncorrected drift" dial on the Totaliser on the Dreyer table -- someone would have a chart down there , presumably, and enter the proper difference between a perfect drift correction and that applied by the imperfect tweak of the inclined sights at the current gun range being used and in this manner the correct final deflection would propagate to the sights at the guns or in the director.

tone
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
tone
Site Admin


Joined: 29 Oct 2004
Posts: 479
Location: Boston

PostPosted: Fri Apr 15, 2005 12:46 pm    Post subject: Reply with quote

I had a setback -- a bad bug that I finally decided was worth reverting 4 days in the code to avoid having to deal with. But many things are now back in place and I'm coming along.

1. Transmitters and receivers are now precisely simulated to use discrete steps, as they did in the RN at this time. Before you can fire the guns, you must order "line up director", "line up receivers" and "line up sights" to prompt the crew to take steps to ensure all devices agree on the data they will relate during the action. They accomplish this exactly as they did in real life.

2. Director now trains in two modes -- training and slewing. Slewing allows fast changes in bearing to be dialled on, but the guns cannot fire during this time as only coarse steps (2 degrees) are sent during the transition. It is precisely as it was done.

tone
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
doobedoo22



Joined: 08 Dec 2005
Posts: 1
Location: UK

PostPosted: Thu Dec 08, 2005 7:08 am    Post subject: Gunnery Reply with quote

Just how many people are going to be needed to 'crew' one of your massively multiplayer ships for a simulated naval combat?

I'm always fascinated and slightly horrified by the mechanical complexity we had to put up with in systems that would, nowadays, all be sorted out by a single application-specific integrated circuit. This doesn't mean I don't admire the realism, of course. But the cost, and reliability implications... wow.

If future naval combats involve electromagnetic pulses, or ECM, we might suddenly have to learn a lot about directing gunfire the hard way again!
Back to top
View user's profile Send private message Visit poster's website AIM Address
tone
Site Admin


Joined: 29 Oct 2004
Posts: 479
Location: Boston

PostPosted: Thu Dec 08, 2005 4:30 pm    Post subject: Re: Gunnery Reply with quote

doobedoo22 wrote:
Just how many people are going to be needed to 'crew' one of your massively multiplayer ships for a simulated naval combat?


Hard to say, of course. In my prototype battlecruiser, I believe I had 50 sailors on board, but most did tasks that would be too uninteresting to attract players very often. The AI was fully capable of controlling sailors the single player the prototype supported was not personally playing.

You could frame a game in any number of ways. If I got going on making a true game of this, it could go to any extreme or somewhere in the middle.

Personally, I think players on a battlecruiser like this would, if they had their choice, take command preferring the following roles in decreasing order of preference when action seemed unlikely:

1. Captain/officer of the deck
2. general unassigned duties on the bridge where watch could be kept and chatter engaged in
3. Signalmen, wireless operators (when activity occurred there)

When action was joined, I'd see the number of duties inviting play growing greatly:

1. Captain (on bridge or in conning tower)
2. Gunnery officer (in gunnery control tower)
3. Spotting officer (in spotting top)
4. Director Layer (he has the trigger!) and Director Trainer
5. Range takers (1-6 of these guys, some in turrets)
6. Range Officer (in spotting top, keeps the range rate)
7. Signalmen (it's FUN tying your vessel into the larger battle)
8. A few men here and there to keep an eye on emerging threats (e.g.: collision, unsafe maneuvering)
9. Transmitting Station Officer (in TS, observes data and suggests range rate)

When torpedo firing or secondary battery fire seemed likely, these would have several duties with great appeal. One drawback to WW-I shooting after director fire was developed was that only a few men on board got the traditionally salable (in the world of computer games) roles of positioning crosshairs on target and pulling the trigger, but the secondary batteries were the last to surrender their local laying (only starting to do so after Jutland, so a typical battleship would have 6-8 more pairs of layers and trainers per broadside once she started to fire her secondaries).

tone
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
Harley



Joined: 23 Oct 2005
Posts: 131
Location: Great Britain

PostPosted: Fri Jan 06, 2006 3:22 pm    Post subject: Reply with quote

Out of interest, and i can't remember reading it elsewhere here on the fora, does the AI work completely in "With the Fleet"? i.e., your robot crewmen will acquire a target, lay the guns, fire ranging shots, correct etc. all on their own, without the player's interference?

Harley
Back to top
View user's profile Send private message AIM Address MSN Messenger
tone
Site Admin


Joined: 29 Oct 2004
Posts: 479
Location: Boston

PostPosted: Fri Jan 06, 2006 4:20 pm    Post subject: Reply with quote

Mostly.

Here are the parts I recall I had to do:

I had to select the target. This was somewhat gamey -- I would click on another ship in the world and then yell down to the TS that we had a target and were going to engage it, e.g.: Target red 50!

The part I did not like in this is that I cheated a bit and made it that this declaration established a "Vulcan mind-meld" mental harmony of what the target was. In part, this made up for my lack of Evershed bearing indicating gear, and also spared me the need to have more sophisticated text parsing logic to deal with disambiguating which ship I meant, e.g.: "No -- the nearer one steaming to the left".

Once I had done that, additional phone men in the TS understood that a target being selected was one of the types of commands which, if they heard it, they were to relay on to the men in the director and at the guns and RFs.

Once this was done, the only other command that I (the human) *had* to issue were to move between major modes of the firing process, such as to open fire and to check fire before "train fore and aft". In the interim, a lot of autonomous dialog would ensue. Not every piece of available direction was exercised by the AI, however. For instance, the spotter would never declare a target inclination angle as was often used at the outset of battle.

An important thing to keep in mind is that the highest level roles had no AI for them at the time I set this aside -- specifically the Captain and Gunnery Officer were not represented. My expectation was that I'd issue these orders (which are few -- essentially selecting a target and opening and closing sustained fire).

By the time I was done, the AI spotting officer knew how to perform ranging logic much as the 1916 spotting rules indicate. The process might overall go like so:

0. (almost forgot this one!) I'd shout down to the TS "line up receivers" and "line up director". This would cause vital synchronization processes to be undertaken. I had no solid feedback on when this process completed, but I know the pointers would be treated well enough that no one attempted to use them before the transmitters and receivers had been brought into agreement). I forget how well the director lining up was done.
1. I'd click on a ship and shout down to the TS, "Target (some damn bearing ... they ignored this data anyhow and relied on the fudgey knowledge of which ship was the target which I endowed them with)".
2. This word percolated to all guns, RFs and directors. If it was nighttime, even the Searchlight man would swivel his lamp to the target and switch it on. That was corny!
3. The range finders would start generating occasional cuts (with errors), and the main RF also generated bearing data.
4. The team in the dreyer would plot this data as it arrived over the receivers, and they'd tune the range to match the first range cut.
5. Before this got too well under way, I'd usually attempt to spice the early solution by calling out an inclination and estimate of the enemy speed, e.g.: "Enemy's course 60 to the left at 20 knots". This would cause the TS people to set the dumaresq accordingly, and then dial on the range and bearing rates implied -- a state of configuration that would persist for a few minutes until such time as the correlations on the plots gained some mettle.
6. Then, I could sit back and relax. The TS would call up occasional range rate suggestions to the Range Officer in the top. He was rather idiotic in his AI, and he'd generally rubber stamp any that had any confidence at all. The longer I let this go on, generally one would expect a truer solution to result. A cheater typing command could be used to compare actual circumstances to the perception shared by the fire control staff.
7. I could then choose to call out single half-salvoes or just declare "open fire". Once a sustained mode of fire was adopted, the process could once again largely go on without my intervention.
8. the Spotting Officer had EXCELLENT AI by the final version -- he'd bring the first shots on for deflection, and then do a bracketing exercise just as outlined in the 1916 spotting rules. His methods for recovering from lost deflection or for an unexpected shotfall were 100% on, but there was not support for double salvoes used in speeding up the initial finding of the range. The Spotting AI was hamstrung in that I did not automatically permit him to see the result properly. That is, the longer the range, the lower the odds that he'd see it at all, and he was prone to miss seeing hits and overs, particularly if there were shorts in the same salvo.

If you see the last "Analysis" video (right click here and save to desktop) I posted just about a month ago, you'll see evidence of how the Spotter AI could fail to see the proper result and how this would screw things up. My part of the dialog shown in this forensic replay are marked as "Gunnery Officer" -- all other chatter is from AI sailors, and much of that is not shown in this! There is a LOT of teamwork going on, and with few exceptions the AI really does listen to what other sailors say whether they are real or virtual.

Of course, in any game, getting the graphics right so that HUMANS can only see the actuality with approximate realism would be a big goal, as would making the AI sailors match this benchmark performance (or do a little less well), or people would complain about the AI eyes having superior ability to human eyes.

I promised I'd make a web page explaining this confusing analysis video in some detail, but I'll cheat and provide a small legend here for the display at the bottom edge.

Time of Day (replay speed)
Range: reality (range plot pencil on Dreyer)
Range Rate: reality (rate on range clock)
Bearing Rate: reality (rate on bearing clock)
Sight settings on director: range in yards, deflection knots

On the overhead map, our ship traces a green wake (to the north, zigzagging eastward), and the enemy steams due east with a black wake to the south. A translucent ship is drawn on the map at a range from our own vessel at a range along the line of bearing equal to the range on the Dreyer's range plot. The course of this ghost ship is sythesized so that he steams down the trend demanded by the given factors: he remains at the range plot pencil reading from our own ship, and he is at the actual relative bearing of the enemy vessel.

A translucent large bullseye near the ghost ship indicates (I think) where we'd expect our shells to fall if we fired them. As such, their deviation from the ghost ship is composed of the following factors: the distance our own ship will travel during the time of flight to the range on the director's sights, and the spotting corrections applied relative to the range plot.

Range cuts are depicted but are hard to make out in this. You'll see a . (for main RF) or a T (for torpedo control tower RF) drawn on the map near the enemy along our line of bearing drawn at the range being reported with a stem connecting this datum to the actual enemy position at the moment it was made. Therefore, the lengths of these stems indicates the errors being made in range cuts.

When the guns are fired, the large bulleye spawns a smaller, blacker bullseye where the salvo might be expected to fall. The settings on the sight at the time the salvo left is noted on the map. You can see this first done at 1:20 into the video.

Spotting corrections and rate commands yelled down from the top are also placed right on the map to slowly fade out. How much more chart junk can we tolerate?! I wish I had voice cues for that.

At 1:43 into the video, you see a salvo arriving. Blue splashes or yellow explosions note each shot's final position, and a stem is then drawn from the bullseye for the salvo to the center of where the target was when the salvo fell. These stems, then, denote the errors in shotfall (you have to guess our bearing at the moment by finding a nearby range cut stem, as these indicate our bearing.

Lastly -- you'll notice that you can actually see range corrections as they are dialled onto the sight in the director. Deflection adjustments are made more coarsely, but you'll see the translucent bullseye move about when these alterations are made.

I am very proud of the graphical tricks used in this sort of display, but confess that it does require some reduction in complexity, or at least provide the means by which you can turn various displays on and off over time.

tone
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
Harley



Joined: 23 Oct 2005
Posts: 131
Location: Great Britain

PostPosted: Mon Jan 09, 2006 7:36 am    Post subject: Reply with quote

I have to say I was very impressed by the display. It's very similar to a much cruder version used to display traffic on Microsoft Train Simulator. Interesting to see most actions displayed so concisely.
Any plans on implementing action from the conning tower, chart room and the flag plot. It would be nice to be able to physically control a squadron or a division or even just your own ship in battle.
Back to top
View user's profile Send private message AIM Address MSN Messenger
tone
Site Admin


Joined: 29 Oct 2004
Posts: 479
Location: Boston

PostPosted: Mon Jan 09, 2006 9:22 am    Post subject: Reply with quote

Of course I'd want those, Harley, but you cannot imagine the degree to which my sim is "on the blocks" at the moment. I am in do-over mode for a long time yet to come, and am a year at least from reconstituting what I had in the prototype, I'm afraid.

A simulation that fails to put ship maneuver into the squadron and fleet level is just visualization or a weapons wank. Indeed, if I had to cut corners during development, I'd axe the ability of the player to play the helmsman at all. Dreadnoughts were not jet-skis which motored blithely about at the flick of a wheel, and focusing on granting a player the means to conduct them so undermines the predominant mode of navigation which would be conformance to formation and station-keeping while an enemy is engaged.

tone
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
Harley



Joined: 23 Oct 2005
Posts: 131
Location: Great Britain

PostPosted: Mon Jan 09, 2006 5:14 pm    Post subject: Reply with quote

I can only imagine, Tone, all the effort you've put into the simulation, and I can see why you would want to cut corners in certain places so as to concentrate elsewhere.
What I meant was not so much being at the wheel of the ship (doesn't really hold much attraction for me anyway) but being OOD and being able to give certain indirect helm commands, or being able to plot a fairly rigid course from the Navigation room (Navigator position, mark course on chart), certainly nothing so cavalier as taking a joyride on a BB. I've read enough books on Destroyer Command to know how very very complicated handling a DD is, let alone a BB.
I hope the reconstruction of the simulation goes relatively smoothly anyway.

Harley
Back to top
View user's profile Send private message AIM Address MSN Messenger
tone
Site Admin


Joined: 29 Oct 2004
Posts: 479
Location: Boston

PostPosted: Mon Jan 09, 2006 6:01 pm    Post subject: Reply with quote

I'd certainly like that have that all in place, Harley.

I did have some reasonable stuff in the prototype, but I also had questions about how exactly the people on the bridge would command a given speed and what indications they had on rudder angle, etc.

Here is what my prototype had.

When YOU were the Helmsman, you could move the wheel back and forth through its range using the arrow keys (shift made the rotation faster), but I did not provide any indication of rudder angle. Additionally, you could select either engine telegraph and move it through its range of commands, and this would affect the speed of the ship. The extent of the realism of the ship's steaming and responsiveness was laughable, really -- your overall maximum speed would be realized when port and starboard were at "ahead full", and each command taken off reduced the speed by 1/6th (full, half, slow --- 3 commands for each of two engines). Similarly, your ship simply turned under helm at a yaw rate equal to (max turn rate * rudder deflection / maximum deflection). Sad, eh?

When your helmsman was in AI mode, he actually did a better job. He could respond to commands in the following fairly numerous variations (typed):

(all | port | starboard) (ahead | back) (full | half | slow)
(all | port | staboard) stop
turn <number> (points | degrees) [to] (port | starboard)
steer <compass degree number> [degrees]
steer <compass point> --- (any of the 32 points of the compass)
maintain this heading
(there were some rudder commands... I forget how many I supported)

and... the last one I added was
zig-zag (1 | 2) points
stop zig-zagging

which would prompt him to zig zag that many points each side of his present heading every 2 minutes or so.... this gave him some autonomy, and helped provide a simple hands-free scenario where either own ship or target maneuvered.

tone
Back to top
View user's profile Send private message Send e-mail Visit poster's website AIM Address
Display posts from previous:   
Post new topic   Reply to topic    The Dreadnought Project Forum Index -> Simulation All times are GMT - 5 Hours
Page 1 of 1

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Powered by phpBB © 2001, 2005 phpBB Group