[REL] AI launched torpedoes - Page 5

keltos01 · 09-03-09, 04:29 PM

Quote:

Originally Posted by DarkFish

It are real torpedoes, no shells. Simply said it's a cannon spawning a torpedo when it fires. So your IJN torpedo mod could still be of use

I haven't tested it but they should fire at subs too as it behaves like a normal gun and thus shoots at everything a gun shoots at.

good ! use my mod as you please then ! glad it can be of use

nice thing : spawning a torp... must think of other things to spawn... airplanes for my Jyunsen B ??? possible ???

keltos

miner1436 · 09-03-09, 04:42 PM

Check out this thread, dunno if it would work but I have some airplane ideas.
http://www.subsim.com/radioroom/showthread.php?t=154868

polyfiller · 09-03-09, 06:15 PM

Right ... more testing tonight and I think I've worked out how to implement torp launches which "lead" the target properly. I have already successfully made turret models which mount on other turrets (to provide Battelships with the AA turrets sometimes mounted on them).

Now this seems irrelevant but ... the problem with the torp launches is if you get the AI to lead the target properly ...you do so by setting a speed on around 17 metres / second in the .sim file ... this causes the turret to rotate in the right way ... but then the torps hardly move. So my idea is to create a new dummy turret which fire invisible, silent shells at the 17 m/s speed - and this will rotate the amount we want. Now we mount the real torp launcher on this turret and restrict it's movement so that it cannot rotate at all. I think the AI will still try to fire ... for example when you set the speed to 17/ms for the launcher it does not have the range to reach the target ... but still insists on firing.

Will try and implement the above tomorrow. Everything else I have tried is a bodge and gives varying results based on range and angle on the bow ... and by variable I mean I can get 100% hits at 2000m 90 deg. AOB and anything else .... virtually nothing.... so the real goal here is to get the AI to estimate the lead on the target correctly.

DarkFish · 09-04-09, 05:54 PM

Quote:

Originally Posted by polyfiller

Right ... more testing tonight and I think I've worked out how to implement torp launches which "lead" the target properly. I have already successfully made turret models which mount on other turrets (to provide Battelships with the AA turrets sometimes mounted on them).

Now this seems irrelevant but ... the problem with the torp launches is if you get the AI to lead the target properly ...you do so by setting a speed on around 17 metres / second in the .sim file ... this causes the turret to rotate in the right way ... but then the torps hardly move. So my idea is to create a new dummy turret which fire invisible, silent shells at the 17 m/s speed - and this will rotate the amount we want. Now we mount the real torp launcher on this turret and restrict it's movement so that it cannot rotate at all. I think the AI will still try to fire ... for example when you set the speed to 17/ms for the launcher it does not have the range to reach the target ... but still insists on firing.

Will try and implement the above tomorrow. Everything else I have tried is a bodge and gives varying results based on range and angle on the bow ... and by variable I mean I can get 100% hits at 2000m 90 deg. AOB and anything else .... virtually nothing.... so the real goal here is to get the AI to estimate the lead on the target correctly.

this might actually work

I've played with setting a low shell speed myself but that would change the elevation angle (which has to be 0 or the torps will run deep/shallow) and the launcher refused to fire so I gave up. I didn't think of adding a launcher on a seperate gun for the train angle. Nice idea.
Please post the results here

polyfiller · 09-07-09, 03:43 AM

OK so mounting the torp launcher on another dummy turret did nothave the desired effect; the AI Knows it can't hit the target if the dummy turret is point at where it wants to, based on a shell speed of around 20 m/s and therefore the torp launcher mounted on it refuses to fire.....

However, there may still be some mileage in this approach - the other method I've used is to rotate the torp launcher 90 degrees so that the AI's use of gun barrel elevation effeectively leads the target. This works OK from some wide angles but at accuate angle the launcher rotates around it's 90 degree axis and therefore gets nowhere near traingin on the target. If however we rotate the torp launcher on a dummy turret which has the same or higher shell speed, the dummy turret bring the torp launcher on to some sort of good line to the target, then the elevation kicks in. Looks like this is going to take some tuning though. On the downside it means on launcher mount is needed for each side facing forwards and same for stern facing etc. For destroyer side mounted four turrets are need (port forward facing, port rear facing and same for starboard) .... because to make use of the elevation, the torp mount has to be rotated 90 degrees different ways.

I'm eve thinking the best bet for a sub is 2 x standard launchers on the front, plus two x dummy turret launchers (one port, one starbd) also up front - this gives good targetting up close from the standards and one good aimed short if the target is leading or behind in AOB terms.

This is NOT easy....

polyfiller · 09-08-09, 04:01 AM

Further update - the dummy turret idea is not working - for some reason the AI can't handle the targetting - the dummy turret itself will rotate to the expected angle to the target - if you vary the shell speed it leads the target more or less relative to shell speed (leads more with slower speeds and points straight ontarget withfaster speeds). The trouble is the poxy AI can NOT correclty aim the turret mounted on the dummy turret - it tries to apply the correct rotation degrees - but the wrong direction - so if the dummy roatets 15 degrees to port to track the target, the mounted turret track 15 degrees to starbord. I hate this game engine

SO I am back to trying two approcahes;

1) enable a spread to be launched - best approach is to tilt the launches on the sub / destroyer - am currently running tilt angles in S3d of -1.57 -0.9 -0.5 0.5 0.9 and 1.57 acroos six launchers onthe front of the sub. Gives a nice spread but many torps wasted. The best shell speed for the above seems to be 200 m/s.

2) Try and force the torp to actually move if we reduce the shell speed to a more realistic 20 m/s and get the AI to aim the turret correctly for the actual torp speed. This is proving the barrel points straight up at low shell speeds regardless of what I put in for range / angle in the .sim. Have tried modifying speed, speed direction and also inherited interia in the particle generator - no joy. Wondered if the global parms would affect it - tried global speed scale - nothing ... this si not looking promising. Help !

DarkFish · 09-08-09, 04:38 PM

Quote:

Originally Posted by polyfiller

Further update - the dummy turret idea is not working - for some reason the AI can't handle the targetting - the dummy turret itself will rotate to the expected angle to the target - if you vary the shell speed it leads the target more or less relative to shell speed (leads more with slower speeds and points straight ontarget withfaster speeds). The trouble is the poxy AI can NOT correclty aim the turret mounted on the dummy turret - it tries to apply the correct rotation degrees - but the wrong direction - so if the dummy roatets 15 degrees to port to track the target, the mounted turret track 15 degrees to starbord. I hate this game engine

SO I am back to trying two approcahes;

1) enable a spread to be launched - best approach is to tilt the launches on the sub / destroyer - am currently running tilt angles in S3d of -1.57 -0.9 -0.5 0.5 0.9 and 1.57 acroos six launchers onthe front of the sub. Gives a nice spread but many torps wasted. The best shell speed for the above seems to be 200 m/s.

2) Try and force the torp to actually move if we reduce the shell speed to a more realistic 20 m/s and get the AI to aim the turret correctly for the actual torp speed. This is proving the barrel points straight up at low shell speeds regardless of what I put in for range / angle in the .sim. Have tried modifying speed, speed direction and also inherited interia in the particle generator - no joy. Wondered if the global parms would affect it - tried global speed scale - nothing ... this si not looking promising. Help !

send your files over to me, and I'll take a look at them. I've got a good understanding of what actually happens if they are fired (since I made them) so maybe I can solve it. And to be honest I think your idea might (should

) work.

(yum... I like tweaking dat files... even better than pizza!

)

keltos01 · 09-08-09, 05:05 PM

Quote:

Originally Posted by DarkFish

(yum... I like tweaking dat files... even better than pizza!

)

keltos

polyfiller · 09-08-09, 05:08 PM

Darkfish - OK ... I'll need to zip up the whole cap ships pack over the wekend ... TBH, I don't think the dummy turret will work ... doesn't matter whether or not the "child" turrent mounted on the dummy turret is launching torps, throwing mackerell or firing normal shells .... it rotates the WRONG direction. It rotates fine right upto the point where you give it a valid barrel name in the .sim file ... i.e. it tracks the target until it knows it's able to shoort properly.... then it goes off in the wrong direction. Strange.

I think the only think that will improve the accruacy is trying to overcome the issue of reducing the shell speed in the sim to around 20 ms and still have the torps come out the tubes and run towards the target... I've played with just about everything in the particle generator and I can get them to run at 90 degree to the intended track real fast, I can get them to bob up & down on the intended track but going nowhere and even flying straight down (destroys the launching sub itself) and backwards ... oh and even the direction the sub is going .... in fact I wouldn't be surprised if I've managed to make them travel in a fourth dimension .... basically any which way apart from the one we want

at a slow enough launch speed for the turret to lead the target.

polyfiller · 09-08-09, 06:43 PM

Quick update on a convoy test mission I've just run .... 12 merchants, 4 escorts (including me in a V&W DD) ... 5 AI controlled U Boats straddling the convoy in classic attack positions. Result - 2 merchants survived plus me.... everything else sunk.
Only problem I am having is to control AI accuracy I am trying to limit the range the AI attempts to fire the torps ... I've set max range in the .sim to 2000m but the furthest sub from the convoy is nearer 4000 at the start of the mission and it insists on firing from that range. Any way to control the AI's understanding of range ?

Bubblehead1980 · 09-09-09, 01:43 AM

Reading these posts about AI controlled torpedos from subs and ships is a big tease(lol) Thanks for the work guys, can't wait till your work is implemented in major mods out there, will be too cool.

DarkFish · 09-09-09, 04:10 AM

Quote:

Originally Posted by polyfiller

Darkfish - OK ... I'll need to zip up the whole cap ships pack over the wekend ... TBH, I don't think the dummy turret will work ... doesn't matter whether or not the "child" turrent mounted on the dummy turret is launching torps, throwing mackerell or firing normal shells .... it rotates the WRONG direction. It rotates fine right upto the point where you give it a valid barrel name in the .sim file ... i.e. it tracks the target until it knows it's able to shoort properly.... then it goes off in the wrong direction. Strange.

believe me, i've seen some strange things. One time I had the player turret rotate into the wrong direction, if you looked to the left the barrel rotated to the right. Solving your problem is just a question of recreating mine. I'm confident there must be a way to invert its rotation.

Quote:

Originally Posted by polyfiller

... doesn't matter whether or not the "child" turrent mounted on the dummy turret is launching torps, throwing mackerell or firing normal shells ....

NOOOoooooooo, it's polyfillers terrible mackerell throwing cannon!

we'll never get the stench of rotten fish off our ship! RETREAAAAAT!!!!

Hitman · 09-09-09, 07:27 AM

Brianstorming here, two ideas for you:

A) If it is finally possible to limit the distance at which the Ai will "shoot" that torpedo? If you can force the AI to shoot only at close distance, like f.e. up to 1000 yards (Or even less), you have already a good torpedo for submarines and which has lots of chances of hitting.

B) What about spreads? Instead of leading the target, you could put the torpedo tubes in fixed position, BUT not aiming at zero degrees. Instead, make them have a spread and give the unit salvoes of three: One goes zero Gyro Angle, the others 5º right and left. Should cover fairly well the area up to 100 metres for any target crossing in front of the submarine.

Basically, I'm thinking about adding enemy submarines that can appear and shoot a salvo at you from periscope depth. If they get close and fire a spread -which is what they actually did in real life- you could have already something very good

polyfiller · 09-09-09, 08:24 AM

Hitman - not sure how I can limit the subs engagement range with the torps .... only way I sey is to reduce the AI's sensor range - but then they could sail right past a target and not move in to engagement range.

Darksifh - any ideas on this one.

As for torp spreads ... yes .. already done this by rotating the launchers. My current best solution has six launchers on the fron of the sub roatated around the horizontal axis - so they are titled to one side. In S3d terms the rotation I'm using is -1.5, -1.0, -0.5, 0.5, 1.0 and 1.5 ... give a very wide spread .... may look at reducing .... especially if I can limit engagement range effectively.

keltos01 · 09-09-09, 01:35 PM

Darkfish : I can hardly wait !!!

about the salvo : could it be according to the type of AI boat ? like the Gato fires 6, Porpoise 4 etc ???

could they fire a stern salvo ?

keltos

09-09-09, 01:43 AM	#71
Bubblehead1980 Navy Seal Join Date: Apr 2009 Location: Florida USA Posts: 6,922 Downloads: 550 Uploads: 42	re Reading these posts about AI controlled torpedos from subs and ships is a big tease(lol) Thanks for the work guys, can't wait till your work is implemented in major mods out there, will be too cool.

09-09-09, 07:27 AM	#73
Hitman Pacific Aces Dev Team Join Date: Sep 2002 Location: Spain Posts: 6,099 Downloads: 109 Uploads: 2	Brianstorming here, two ideas for you: A) If it is finally possible to limit the distance at which the Ai will "shoot" that torpedo? If you can force the AI to shoot only at close distance, like f.e. up to 1000 yards (Or even less), you have already a good torpedo for submarines and which has lots of chances of hitting. B) What about spreads? Instead of leading the target, you could put the torpedo tubes in fixed position, BUT not aiming at zero degrees. Instead, make them have a spread and give the unit salvoes of three: One goes zero Gyro Angle, the others 5º right and left. Should cover fairly well the area up to 100 metres for any target crossing in front of the submarine. Basically, I'm thinking about adding enemy submarines that can appear and shoot a salvo at you from periscope depth. If they get close and fire a spread -which is what they actually did in real life- you could have already something very good __________________ One day I will return to sea ...

09-09-09, 01:35 PM	#75
keltos01 Silent Hunter Join Date: Feb 2008 Location: Milan Italy Posts: 4,999 Downloads: 114 Uploads: 18	Darkfish : I can hardly wait !!! about the salvo : could it be according to the type of AI boat ? like the Gato fires 6, Porpoise 4 etc ??? could they fire a stern salvo ? keltos __________________ "Honorable Builder of Sinking Ships"

09-03-09, 04:42 PM	#62
miner1436 Grey Wolf Join Date: Jul 2007 Posts: 808 Downloads: 107 Uploads: 3	Check out this thread, dunno if it would work but I have some airplane ideas. http://www.subsim.com/radioroom/showthread.php?t=154868

09-03-09, 06:15 PM	#63
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	Right ... more testing tonight and I think I've worked out how to implement torp launches which "lead" the target properly. I have already successfully made turret models which mount on other turrets (to provide Battelships with the AA turrets sometimes mounted on them). Now this seems irrelevant but ... the problem with the torp launches is if you get the AI to lead the target properly ...you do so by setting a speed on around 17 metres / second in the .sim file ... this causes the turret to rotate in the right way ... but then the torps hardly move. So my idea is to create a new dummy turret which fire invisible, silent shells at the 17 m/s speed - and this will rotate the amount we want. Now we mount the real torp launcher on this turret and restrict it's movement so that it cannot rotate at all. I think the AI will still try to fire ... for example when you set the speed to 17/ms for the launcher it does not have the range to reach the target ... but still insists on firing. Will try and implement the above tomorrow. Everything else I have tried is a bodge and gives varying results based on range and angle on the bow ... and by variable I mean I can get 100% hits at 2000m 90 deg. AOB and anything else .... virtually nothing.... so the real goal here is to get the AI to estimate the lead on the target correctly.

09-07-09, 03:43 AM	#65
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	OK so mounting the torp launcher on another dummy turret did nothave the desired effect; the AI Knows it can't hit the target if the dummy turret is point at where it wants to, based on a shell speed of around 20 m/s and therefore the torp launcher mounted on it refuses to fire..... However, there may still be some mileage in this approach - the other method I've used is to rotate the torp launcher 90 degrees so that the AI's use of gun barrel elevation effeectively leads the target. This works OK from some wide angles but at accuate angle the launcher rotates around it's 90 degree axis and therefore gets nowhere near traingin on the target. If however we rotate the torp launcher on a dummy turret which has the same or higher shell speed, the dummy turret bring the torp launcher on to some sort of good line to the target, then the elevation kicks in. Looks like this is going to take some tuning though. On the downside it means on launcher mount is needed for each side facing forwards and same for stern facing etc. For destroyer side mounted four turrets are need (port forward facing, port rear facing and same for starboard) .... because to make use of the elevation, the torp mount has to be rotated 90 degrees different ways. I'm eve thinking the best bet for a sub is 2 x standard launchers on the front, plus two x dummy turret launchers (one port, one starbd) also up front - this gives good targetting up close from the standards and one good aimed short if the target is leading or behind in AOB terms. This is NOT easy....

09-08-09, 04:01 AM	#66
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	Further update - the dummy turret idea is not working - for some reason the AI can't handle the targetting - the dummy turret itself will rotate to the expected angle to the target - if you vary the shell speed it leads the target more or less relative to shell speed (leads more with slower speeds and points straight ontarget withfaster speeds). The trouble is the poxy AI can NOT correclty aim the turret mounted on the dummy turret - it tries to apply the correct rotation degrees - but the wrong direction - so if the dummy roatets 15 degrees to port to track the target, the mounted turret track 15 degrees to starbord. I hate this game engine SO I am back to trying two approcahes; 1) enable a spread to be launched - best approach is to tilt the launches on the sub / destroyer - am currently running tilt angles in S3d of -1.57 -0.9 -0.5 0.5 0.9 and 1.57 acroos six launchers onthe front of the sub. Gives a nice spread but many torps wasted. The best shell speed for the above seems to be 200 m/s. 2) Try and force the torp to actually move if we reduce the shell speed to a more realistic 20 m/s and get the AI to aim the turret correctly for the actual torp speed. This is proving the barrel points straight up at low shell speeds regardless of what I put in for range / angle in the .sim. Have tried modifying speed, speed direction and also inherited interia in the particle generator - no joy. Wondered if the global parms would affect it - tried global speed scale - nothing ... this si not looking promising. Help !

09-08-09, 05:08 PM	#69
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	Darkfish - OK ... I'll need to zip up the whole cap ships pack over the wekend ... TBH, I don't think the dummy turret will work ... doesn't matter whether or not the "child" turrent mounted on the dummy turret is launching torps, throwing mackerell or firing normal shells .... it rotates the WRONG direction. It rotates fine right upto the point where you give it a valid barrel name in the .sim file ... i.e. it tracks the target until it knows it's able to shoort properly.... then it goes off in the wrong direction. Strange. I think the only think that will improve the accruacy is trying to overcome the issue of reducing the shell speed in the sim to around 20 ms and still have the torps come out the tubes and run towards the target... I've played with just about everything in the particle generator and I can get them to run at 90 degree to the intended track real fast, I can get them to bob up & down on the intended track but going nowhere and even flying straight down (destroys the launching sub itself) and backwards ... oh and even the direction the sub is going .... in fact I wouldn't be surprised if I've managed to make them travel in a fourth dimension .... basically any which way apart from the one we want at a slow enough launch speed for the turret to lead the target.

09-08-09, 06:43 PM	#70
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	Quick update on a convoy test mission I've just run .... 12 merchants, 4 escorts (including me in a V&W DD) ... 5 AI controlled U Boats straddling the convoy in classic attack positions. Result - 2 merchants survived plus me.... everything else sunk. Only problem I am having is to control AI accuracy I am trying to limit the range the AI attempts to fire the torps ... I've set max range in the .sim to 2000m but the furthest sub from the convoy is nearer 4000 at the start of the mission and it insists on firing from that range. Any way to control the AI's understanding of range ?

09-09-09, 08:24 AM	#74
polyfiller Weps Join Date: Sep 2005 Posts: 356 Downloads: 40 Uploads: 0	Hitman - not sure how I can limit the subs engagement range with the torps .... only way I sey is to reduce the AI's sensor range - but then they could sail right past a target and not move in to engagement range. Darksifh - any ideas on this one. As for torp spreads ... yes .. already done this by rotating the launchers. My current best solution has six launchers on the fron of the sub roatated around the horizontal axis - so they are titled to one side. In S3d terms the rotation I'm using is -1.5, -1.0, -0.5, 0.5, 1.0 and 1.5 ... give a very wide spread .... may look at reducing .... especially if I can limit engagement range effectively.