Aces High Bulletin Board
Special Events Forums => Friday Squad Operations => Topic started by: Nefarious on July 02, 2020, 10:45:57 AM
-
Would you be for or against using AI bomber formations seen in the Fortress Europe Missions in FSO?
My idea would see AI bomber formations (bombers only) in the FSO environment, especially events that call for heavy or medium bombers. This would be especially helpful for unpopular bomber roles like the Betty, He 111, and B-25C.
Please let me know below.
-
Willing to give it a try.
-
Interesting but how would a CIC control the route or would that be possible?
:cool:
-
Interesting but how would a CIC control the route or would that be possible?
:cool:
Not dynamically, but command could have a menu of missions to choose from, and decide when and in what order to run each.
But once started, it would have to run it's defined course.
-
Interesting but how would a CIC control the route or would that be possible?
:cool:
Not dynamically, but command could have a menu of missions to choose from, and decide when and in what order to run each.
But once started, it would have to run it's defined course.
Trips is right, There is no "controlling" the formation. Only setting a specific path.
In FSO we could do it one of two ways.
1) The Admin CM (me) could make and set the path, making it different every frame.
2) The CIC could create the path, and I would create the mission to their specifications.
In its first running, I would go with option 1. Just because FSO orders seem to run late, and time constraints on my end may cause issues.
-
I am very much not in favor of AI bombers. Their defensive fire is much too strong.
-
I am very much not in favor of AI bombers. Their defensive fire is much too strong.
Currently it is set "randomly" in formations, it could be dialed back even more.
I put randomly in quotations, because it's not really a random setting, it's just that some bombers are better than others in our current formations, but none are maxed out. If I remember correctly, and Trips would know... Our bombers are medium to low for lethality.
-
I am very much not in favor of AI bombers. Their defensive fire is much too strong.
It never seemed that way to me from escorting them in the Fortress Europe missions. :D
To some extent, the concentration of AI defensive fire can be adjusted in the mission definition.
Is that your only concern?
-
What happens when the bombers get to target? Are they surgical with their bomb drops? Do they just drop in the middle of the runway?
The only downside I can see is you can't pressure an AI as it's about to drop, or distract it from being in the bombsight to gun. Its gunnery will also be more consistent across the board than people.
I kind of like the idea of having more fighters in the air though. On the other hand, the few missions I've run against AI bombers they chewed the heck out of me in a couple passes.
I'm not vehemently opposed to it, just curious how the details would work out.
Wiley.
-
I am 100% against AI anything. I understand its utility, but we do not need it in any capacity.
-
I think it would be worth an "experiment" with feedback through the process. Then decide if it is something worth adding in 2021 designs.
-
I guess controlling was the wrong word to use. What I meant was would it be possible to send the CM way points for a route they wanted the AI formation to follow?
:cool:
-
What happens when the bombers get to target? Are they surgical with their bomb drops? Do they just drop in the middle of the runway?
The only downside I can see is you can't pressure an AI as it's about to drop, or distract it from being in the bombsight to gun. Its gunnery will also be more consistent across the board than people.
I kind of like the idea of having more fighters in the air though. On the other hand, the few missions I've run against AI bombers they chewed the heck out of me in a couple passes.
I'm not vehemently opposed to it, just curious how the details would work out.
Wiley.
The bombers operate more like carpet bombers than pin-point smart bombers. I select a point on the map and they drop their bombs. Because of their formation, the bombs fall generally in the same pattern, so I'm not quite sure they even use a bomb sight like humans do.
In the last mission, the AI destroyed 29 objects, While Human bombers destroyed 47. The total object count at the target was 124 objects at the troop strat target.
-
I guess controlling was the wrong word to use. What I meant was would it be possible to send the CM way points for a route they wanted the AI formation to follow?
:cool:
I understood what you meant.
Think of the mission as a predefined set of waypoints with a target speed and alt and behavior associated with each transit to that waypoint.
Once you kick it off, the AI are going to follow those waypoints at the specified alt and speed and drop where you tell them.
Before it can be uploaded and run, it has to be "compiled" and after that can't be modified without recompiling and re-uploading a new one.
So, because it must be compiled by an offline tool, it's really not feasible to define one on the fly in arena or alter one after it's been uploaded to the server.
:salute
-
it's really not feasible to define one on the fly in arena or alter one after it's been uploaded to the server.
Yeah it would be a no change, no matter what, in the orders once sent to a CM.
The bombers operate more like carpet bombers than pin-point smart bombers.
Also good to know
:salute
-
It never seemed that way to me from escorting them in the Fortress Europe missions. :D
To some extent, the concentration of AI defensive fire can be adjusted in the mission definition.
Is that your only concern?
It seemed to me that most pilots attacking the bombers were critically damaged after one or two passes. It also only seemed like only in the practice frames with late war planes did the Luftwaffe pilots have success downing a significant number of bombers.
There are many individual problems with the AI bombers, I doubt that all of them can be corrected to where their survival and kill rates mirror the average survival and kill rates for player bombers.
1. The AI formations were packed much tighter than a group of player flown formations.
2. The guns are too accurate at long range. Most of the times I was knocked out was from a range of about 1000 yards and even farther.
This next issue I have not witnessed in a Fortress Europe but I did witness in AI testing in the AvA. I question if it has been fixed.
3. Bombers with missing wings continued to fly straight and level after losing wings for a significant time and could still provide accurate defensive fire until final destruction.
There is also the issues in how to implement the the AI bombers. Are they going to replace player flown bombers or just supplement a player flown force?
How many AI formations will be used? I see it as being way too easy in going overboard on their numbers. Especially since it's easy for a designer to think that because an event historically had hundreds of bombers participating that a few dozen sets should be fine. For example, the last Fortress Europe event had 43 formations of bombers, 37 of which were AI. That is way too many for any event to be properly playable.
In FSO, a 20 player formation it is almost too many to handle for a defense. The typical engagement at a single objective is with forces of 30 players a side. A designer still needs to work from a base point of having 10-16 formations if he has any hope of having a balanced event - and that's assuming that the AI can be dialed into the suriviability and lethality of average human players.
I also question how effective the AI bombers are at actually hitting their targets. They can't be told to bomb specific structures at a base. A town is too small to provide enough targets for 12-16 formations as only a handful of planes will do the job. I've seen too many bad events rely on defenders needing a ridiculous number of bomber kills to break even on points. Using AI bombers seems to be inviting that sort of disaster. So I only see AI being useful for attacks on strats and cities.
I personally don't see much upside in having them, but I'd be willing to try it out under the right circumstances.
-
I believe in the AI bomber asset for Special Events.
1) They are best as decoys.
2) Defenders see bases flash for recon assignment.
3) Defenders have to ascertain whether the strike group is player or AI.
4) Defenders have to choose to attack AI or resume hunting for player strike group.
I also believe on keeping them attacking only non valid targets. AI targets can be close to player assigned valid targets.
Defenders have the option to keep a recon bird trailing the group for T + 60 action once the player strike groups have hit the valid targets.
-
I'll just throw in some thoughts based on my knowledge of the mission system. I'm not involved with this. I helped Nefarious setup some of the initial FE test missions, but he is is fully in the saddle and making his own missions now. This is his baby, but I'll just offer $0.02.
1. It is almost certainly true the AI formations are tighter than a formation of humans.
We based the formation and spacing to the best of our ability from historical information, but it is probably true that was an overly perfect ideal that real life pilots never attained either.
It would take work, but those formations can be shape spacing anyway that is decide.
2. Yeah. I was only at some of those early FE missions. Not all of it has to do with the AI. A lot of the differences in results had to do with when and where the LW were allowed to airspawn. But no doubt, there proper balancing of variables would be necessary.
Just for background info....
The definition for the AI flight of bombers has a variable, that if I understood correctly, essentially defines the angle of the cone of dispersion of their fire. You can set it small and they might not hit you as often, but when they do you are getting sliced by a laser beam. You can set it larger and they probably have better odds of pinging you, but you are less likely to get sliced in half.
The min and max of this value can be set between 0 and 15. If you set a range, each AI bomber is given a random selection from that range. So the wider the range the more variability between bombers. I can't remember what we used in the early FE missions. Something like 3-7.
We did some testing ourselves in a Custom Arena and just picked what felt reasonable. There is a fair bit of adjustment possible there though. Of course a lot of what I saw in the first FE missions were individual attackers zooming in alone and facing the combine fire alone. Of course that isn't going to turn out well. The importance of teamwork is that it gives the enemy someone else to shoot at. If there was interest, some tests could be setup and try different accuracy values to pick something that felt fair. We just had to make our best guess. But it should be a test with 3 or 4 simultaneous attackers to be meaningful, IMHO.
3. I hadn't seen that behavior. Maybe Nefarious has. It may have been fixed. It may be avoided because we do our formations in a unique way where we are not using the standard HTC flight/drone model. I suspect that may have been drone behavior which we are avoiding. I dunno about that one.
4. That's Nef's call, but I'd agree that it seems reasonable to balance the number of bombers to the number of interceptors. That may not have always occurred in the early FE missions because we were intentionally trying to push the envelope and see what the maximum number of AI bombers we could use. We were trying to find our edge conditions.
It's a lot easier to remove bombers than it was to add them. So formation can be designed large to have as a reusable resource, and then trimmed down as needed for a particular case.
5. The effectiveness of the bombers will be a function of how many you have in the formation and how tight the formation is. They drop in formation, so you can get a visual impression of the dispersion by doing a in air fly by. Actual impact dispersion will be even more dispersed than the formation due to random aerodynamic effects on the falling ord. I could run a test and take a screenshot of various impact patterns.
6. I don't have an opinion on the desirability of including them in your FSO. I am just a technical consultant. ;)
:salute
-
Unless we harness the resource that AI bombers presents, we may never have another DGS. We're going to need to embrace the concept and see what it can offer. We want more scenarios and FSOs that are bomber missions with escorts. AI is how we make those events epic.
Come on over the water is and will be fine...
-
For EW setups, imagine 50 sets of Bostons or Ju-88s. FSO Frame Objectives has all Allied and Axis strictly in fighters. One part of a Frame Objective is to escort the AI bombers to target.
Players don't have to complain about flying death traps no more. The AI takes the burden of poorly defendable bombers.
The same can go to the Pacific setups with D3A1s, B5N2s, TBMs and SBDs when strictly in torpedo and dive bombing mode.
Imagination is the blueprint. Adaptability is on us.
:salute
-
Some thoughts on AI:
I'd prefer it to be used in a supplemental form rather than a principal form.
I personally am 100% against AI as I do not think it is necessary at this point in time. Bomber duty is something that FSO squads have endured for nearly 15 years.
AI should be treated as 'auto guns' in an event. Nobody enjoys being shot down or taken out of the fight by manned guns or auto ack, which is why we disable manned ack and set auto ack to a very low damage level (.3 is generally the norm).
I personally would find no enjoyment in finding a large formation of bombers and shooting at them, knowing there are not human players on the other end.
They're pretty, they look cool to dive in on, but are a huge turn-off for the realism enthusiast who plays the game for PvP.
-
I personally would find no enjoyment in finding a large formation of bombers and shooting at them, knowing there are not human players on the other end.
They're pretty, they look cool to dive in on, but are a huge turn-off for the realism enthusiast who plays the game for PvP.
I get what you are saying. I told Nefarious on the first day he asked me to help him get the AI set up for FE, that is the reaction he was most likely going to find in the community.
I don't fly the FSO so it doesn't much matter to me, but let me play Devil's Advocate just to advance the discussion.
First, at least in FE it wasn't all AI. Does it make a difference that some of them are human pilots, you just don't know which ones?
Second, at least in FE, as humans bomber pilots were shot down, Nefarious was inserting them in to take over AI bombers in flight. Replacing AI with humans. So as the mission progressed, a higher and higher percentage of the formation was human. At some point all the remaining bombers might be human. Does that make a difference?
When you intercept a human bomber pilot, do you find it unacceptable that not every gun position on the bomber is manned by a human? The other guns are a form of AI being guided by the actions of the human gunner. Should only the one gun the human is occupying be the only one allowed to shoot? Those AI gunners are augmenting the human gunner the same way the AI bombers would augment the human bomber pilots.
I don't know if they allow formations in FSO, but would it be unacceptable that in a 3 plane formation, only one of those bombers at a time is piloted by a human. The FE AI formations are just an extension of that pattern. A formation of bombers, some human piloted, others AI piloted. As you destroy bombers, the human keeps moving to another bomber until the last bomber is only human. If that is acceptable for a 2 AI: 1 human formation, couldn't it be acceptable for a 20 AI: 6 human formation? As you whittle it down it becomes a higher percentage human formation over time.
When a human bomber drops eggs on a factory, does it matter those are only unpopulated objects and there are no players in those structures they are destroying? Or are they merely a target resource they attempting to reach and destroy while human interceptors attempt to stop them? Could the AI bombers be seen the same way a human bomber pilot see's the factory buildings? Merely a target resource you are trying to reach and destroy while human escorts try and stop you?
Just morsels of food for thought.
:salute
-
Some thoughts on AI:
I'd prefer it to be used in a supplemental form rather than a principal form.
I personally am 100% against AI as I do not think it is necessary at this point in time. Bomber duty is something that FSO squads have endured for nearly 15 years.
AI should be treated as 'auto guns' in an event. Nobody enjoys being shot down or taken out of the fight by manned guns or auto ack, which is why we disable manned ack and set auto ack to a very low damage level (.3 is generally the norm).
I personally would find no enjoyment in finding a large formation of bombers and shooting at them, knowing there are not human players on the other end.
They're pretty, they look cool to dive in on, but are a huge turn-off for the realism enthusiast who plays the game for PvP.
Spikes,
You used the word "endured". Did you mean to use that word?
Endured means that I have to sit in that crate and look down the bomb sight while I hear bullets raddling through my wings and fuselage. OH CRAP! I'm on fire. OH CRAP! I'm in the tower. Nights over...
We can begin to end that with AI in certain events. Who cares whose in the bomber you just blew up.
AI has always been the future and the future is now here.
-
AI I',m not convinced.. ruined the Fortress Europe event I took part in.
The challenge and attraction of AcesHigh is combat against human opposition, and I would expect the reason people pay a subscription or forego other entertainment for SE/FSO.
You can't play mind games with or spook one's and zero's into mistakes or self preservation, there's no soul in it.
FSO appeared to be headed in the right direction, AI implementation is not going to be a step forward the more I think about it.
-
I think it would be worth an "experiment" with feedback through the process. Then decide if it is something worth adding in 2021 designs.
This makes sense to me. With some type of limited usage in actual events and useful feedback, perhaps HT could be persuaded to further develope the AI so that they could add something worthwhile to the events.
-
Just to fill in some of the questions brought up...
I altered one of the formations to spread it out more. It is a 12 plane formation. Dropping from 20k with that formation, here is what the impact dispersion looked like:
(https://www.dropbox.com/s/sm4iw5nyyttdvkn/BDA.png?raw=1)
The white cross-hair is what I set as the target. About 3-4 bombs hit the hangar but don't show as craters.
Wiley, Devil505, (or any one else interested), I just uploaded an offline mission with the more spread out formation, and I set the gunnery to it's lowest settings. There will also be some AI interceptors to help share the defensive fire. Would you mind giving this a try and tell me if the fire is still too accurate? I think it is too low accuracy, but I want you to see the adjustment range available to give you an idea.
You'll want to pick the first flight of 109G6 or 190-5 so you put close to the bombers right away. The g6 have gondolas. The AI interceptors are pretty stupid, but they help draw off some fire.
Look for "Easier Bomber Intercept" in Offline Mission.
Let me know what you think, please.
:salute
-
I flew it 3 times.
Against 12 solo bombers I was able to make 8 passes before the AI fighters killed all the bombers. I would kill 2 on average with 2-3 more assists.
I would receive a hit from defensive fire (hit noise played) once per attack at least 6 times per mission - usually only one hit though. I would lose a part about half the time I took a hit. Based on the tracers from the bombers and assuming that I was indeed only being hit every time the hit noise was played, there were hundreds of near misses. It looked like I should have been hit many more times than I apparently was.
Other observations:
1: I shot the right wing off a B-17. It immediately lost control and stopped firing it's guns.
2: I set a B-17 on fire. It continued shooting until it exploded.
3: A single bomber could fire in multiple directions at the same time.
4: Bombers did not always fire at the closest enemy. I was being shot at by a bomber across the formation while closer bombers were not firing.
5: Bombers would occasionally fire in directions where no enemies were.
6: Bombers opened up firing at a wide variance of ranges. I observed as far as 2.5K and as close as 800 yards. Most of the time it was about 1.5K.
-
3: A single bomber could fire in multiple directions at the same time.
4: Bombers did not always fire at the closest enemy. I was being shot at by a bomber across the formation while closer bombers were not firing.
5: Bombers would occasionally fire in directions where no enemies were.
6: Bombers opened up firing at a wide variance of ranges. I observed as far as 2.5K and as close as 800 yards. Most of the time it was about 1.5K.
We've seen some of that odd shooting off into space. They are just jumpy I guess. ;)
I can create the size and shape of the formation and adjust the dispersion of the firing cone. That is pretty much all I can control other than some flight dynamics.
So previously you said the AI defensive fire was too strong. Did it still seem too strong with those settings and that formation?
:salute
-
Hard to say if what you have is any better. I was attacking 12 solo bombers. In Fortress Europe I was attacking 43 sets of bombers.
I was losing a part every 2 passes against 12 bombers but could stay in the fight. In FE I was either dead or near death after 2 passes. From what I saw by tracer fire in your test mission, I should have been dead or dying.
In either case, it is far more dangerous than attacking a formation of 12 player controlled sets. That is bad.
If a designer wants an event to have AI, the bots should not ever be on average better than a player.
-
Hard to say if what you have is any better. I was attacking 12 solo bombers. In Fortress Europe I was attacking 43 sets of bombers.
Well, that is sorta the point. I think part of the idea would be to balance the number of bombers to some reasonable percentage of the attacking interceptors.
I was losing a part every 2 passes against 12 bombers but could stay in the fight. In FE I was either dead or near death after 2 passes. From what I saw by tracer fire in your test mission, I should have been dead or dying.
In either case, it is far more dangerous than attacking a formation of 12 player controlled sets. That is bad.
Really? It didn't seem that way to me. I regularly get shredded attacking a single 3 plan human formation. I was dancing through these twelve almost at will. I got pinged a couple of times, but I don't think I even lost a part. Maybe I was just lucky.
So you and a couple of brain-dead AI interceptors wiped out a formation of 12 bombers with a combined defensive fire of 144 .50 cal guns, and you killed them all and lived through it. :aok
Anyway, thanks for trying it.
:salute
-
Like I said before, your test does not adequately represent what is typical in an event, FSO or otherwise.
Rarely are single bombers ever used by players, but when they are they die just as easily if found unescorte. But it's foolish to grade the effectiveness of the AI when the result is the same as a bad player plan gone poorly.
BTW those "Brain-dead bozos" shoot better than most players from the looks of it.
-
Like I said before, your test does not adequately represent what is typical in an event, FSO or otherwise.
No, that wasn't what I was trying to do. I just wanted to demonstrate the difference in leathality between the settings we normally use and the lowest settings. It just didn't seem to make sense to have all 12 bombers shooting just at you.
I just uploaded another one called "Harder Bomber Intercept" that is an exact duplicate but with the hardest settings. Having those two to compare is probably better demonstrate the difference in the settings effects.
Anyway, I was just trying to show an example of what could be adjusted.
:salute
-
Some thoughts on AI:
I'd prefer it to be used in a supplemental form rather than a principal form.
I personally am 100% against AI as I do not think it is necessary at this point in time. Bomber duty is something that FSO squads have endured for nearly 15 years.
AI should be treated as 'auto guns' in an event. Nobody enjoys being shot down or taken out of the fight by manned guns or auto ack, which is why we disable manned ack and set auto ack to a very low damage level (.3 is generally the norm).
I personally would find no enjoyment in finding a large formation of bombers and shooting at them, knowing there are not human players on the other end.
They're pretty, they look cool to dive in on, but are a huge turn-off for the realism enthusiast who plays the game for PvP.
Endured? Some of us would rather fly bombers. I cant afford the upper scale computer set ups that can get over a hundred frames per second.
-
Unfortunately bomber duty is often something to be "endured" in the FSO. In a US heavy for example; you expect one guy to man an aircraft designed to be operated by a crew of ten. Granted not every one of those ten crew positions are required to fly a simulated run. But at some point you have to commit to the bombadier position and from then until the end of the run you're basically defenseless. It's not very realistic of what attacking a heavy bomber would actually be like. After 15 years of doing it this way, just maybe it's time to explore other alternatives.
-
no, i cant handle the 'AI on my system,
-
Endured? Some of us would rather fly bombers. I cant afford the upper scale computer set ups that can get over a hundred frames per second.
AH III does not require anything close to a high end machine.
-
AH III does not require anything close to a high end machine.
Good info. :aok
-
Not a fan of AI, willing to try it but I really don't see why it would be needed.