AI landings again

Squiffy

Squiffy

Charter Member
Since I am working on landings in DPC and have worked on some air files, I couldn't help but wonder about AI landing quality. When working with the F9 Panther, I noticed that even at an optimum glide speed with full flaps (player controls) set up for AI, the planes still porpoise up and down as if the pilot skill is way off and the computer is unable to hold a good glide attitude. Now I know that the flap settings in the panther are rather sensitive. This got me thinking that perhaps there is something with the flap settings that the ai fm is not accouting for, or is exaggerating too much. I looked at skill settings in the mission file I built, and turned them up to 3 or 4. No effect. I remember clearly seeing the flap settings in Air Wrench and I wonder if there is something there we can work on to help smooth out the approach. I will look at some now and see if anything turns up. When looking at the models in game cntrl-shift-F4, I see full flaps and wicked porpoising. Bang endo nose wheel crash.
 
OK Lewis, the last one of four landed. The FM does seem a little slippery though. With all the laundry hanging out and only a slight glide slope I was picking up speed pretty easliy. The speed brake didnt' seem too effective and the flap damage speed was a little too low. You should be able to land hot with some flaps like combat flaps or more. Think of an old warbird doing a two wheel landing. Anyways, this is a carrier jet so defintely different. I'll look at the fm now.
 
Squiffy, any headway at all w/ this'll make ya a hero... A.I. piloting skills in cfs3..... whew! way worse than awful--
...all them missions, arriving home-- victory--even survival-- so hard fought, just to watch my wingmen pack it into a dirt mound or a tree... bye-bye emersion:faint:
 
Big Fat Grumman Landing Gear!

I hear ya Popsaka! Now check this out... On Lewis' lead, I adjusted the flap pitch and drag effects down to a minimal aoa value. I even gave a bit nose up just to see. Porpoising is still there, it's in the dice or hard code for the ai. I wonder if the "flight director" settign in the air file woudl make a difference? So, then I beefed up the langing gear contacts damage for carrier duty. I went over the top here. 4000ft/sec nose, 6000 mains. They can come in on the nosewheel, bounce, stall, pancake down and roll away like a dirt bike. Pretty funny to watch! Now the object settings for the routes on the carrier airbase help like in ETO. Longer runways, fat ground squares, and concrete surface. I taxied over to the destroyer to give them some room. :hand: Weight and balance is still the same as Mudpond. Those figures are very close to historical. ChecK out these screenies.
 
AI Landings redux

Hi Squiffy, the quality of the FM really seems to make a difference. I remember when first flying MAW, how much better the AI behaved when making a landing.

A big factor is the terrain - if the airfield is set into the landscape, or perched on a hillside, then the AI seem to fly to what the "iriginal" terrain is, and hence they plough into the ground. What a pain, like Pops says! :argue:

The beautiful Beau is a great plane for AI landing success..
cheers,
D
 
Nice shots. Maybe we should look at how the .air files are arranged for those that have a nice clean glide slope. Adjusting the contacts poitns helped a lot, just in survivability, not in glide quality. But the glide directly effects the ability to survive touchdown. The panther now noses in, bounces way too high chops the throttle and then pancakes down. The heavy gear helps some but I still get a couple explosions now. I think it's my wingman who isn't as "good" as the other flight or some nonsense. Anyway, the flap and gear settings do seems to have a slight effect for pitch but the all around glide slope is the same stupid porpoising.
 
Squiffy said:
...Maybe we should look at how the .air files are arranged for those that have a nice clean glide slope....
Click to expand...
Sounds like a good plan. Always worth comparing files from those that work, with those types of similar a/c that don't land so well.
 
Hope I'm not stating the obvious, but it seems different airfields dictate landing behavior w/ ai (outside of the the way trees etc are not noticed)
 
I noticed that with my P-55s, anytime I had the heaviest load of fuel and rockets, the AI would porpoise in formation, sometimes 100's of ft in altitude up and down.

I fixed it by moving the CoG (in the CFG file)from "0", foward to 0.27 (Stock CFS3 P-55 is 0.54) and it fixed the problem. The AI now can fly straight and level with me.

This isn't landing related, but maybe your FM has the CoG in a bad spot.
 
Old Crow yo could be on to something! You actually got this change by editing the air file? Maybe what we have here is a little better view of the ai. I tseems it is limited in response time and can't keep up with more squirrely stability balances. The COG change might just do the trick. I had my fuel set to 1% or very low. No effect. But... I think the F9 is already at zero. Lets keep at it.
 
OldCrow said:
I noticed that with my P-55s, anytime I had the heaviest load of fuel and rockets, the AI would porpoise in formation, sometimes 100's of ft in altitude up and down.

I fixed it by moving the CoG (in the CFG file)from "0", foward to 0.27 (Stock CFS3 P-55 is 0.54) and it fixed the problem. The AI now can fly straight and level with me.

This isn't landing related, but maybe your FM has the CoG in a bad spot.
Click to expand...

...so simple its brilliant :ernae:
 
Squiffy said:
Old Crow yo could be on to something! You actually got this change by editing the air file? Maybe what we have here is a little better view of the ai. I tseems it is limited in response time and can't keep up with more squirrely stability balances. The COG change might just do the trick. I had my fuel set to 1% or very low. No effect. But... I think the F9 is already at zero. Lets keep at it.
Click to expand...
No Squiffy, it was in the CFG file. A positive is foward, and a negative is back I believe.

looks like this...


reference_datum_position = 0, 0, 0
empty_weight_CG_position = 0.27, 0, 0
 
OK, well they are connected so I will work and save in airwrench just to be sure. If AI lacks trim, this could be helpful in developing a workaround for ai flight control in all configurations.
 
Squiffy said:
OK, well they are connected so I will work and save in airwrench just to be sure. If AI lacks trim, this could be helpful in developing a workaround for ai flight control in all configurations.
Click to expand...
Where does the CoG factor in on the AIR file? I think it is in record #1101?

How does it equate to the CFG file?
 
Just a bit of food for thought, gents.

Your talking about the AI flying abilitys.

Back in the CFS2 days, I ran into some of it, asked questions, and ran into brick walls. But...it does lead into the FM, both the airfile and cfg, since they work concurrently.

MS would NOT release any hard info, since it would dig into 'their' FM trade secrets.

I'm fairly certain, though, one unknown entry that can overall affect the AI, is the airframe type number.

Secondly, fairly critical for the AI, is weight and balance. Realistic weights to control much of the flying characteristics.
Balance, for the sim and the AI, needs to be very close, even if a little adjusted.

Looking at the AI realisticlly, consider the 'porpoising thing' for example. The AI has it's limits, which are a little slower in actions. If balance is off, the a/c tends to nose up, or down, and the AI spends it's time trying to correct for level flight. There's your porpoising.

For the AI, things need to be as accurate as possible.
 
I completely understand crossram. that all makes sense. maybe the pilot weight is off? I looked at the weight and balance. Also MOIs on the stock Panther were not updated. I tested them with updates. more slugish control at low speed and maybe only a hair less porpoising. I don't minid it so much now that I have stornger landing gear to work with. It' sstill kind of fake anyway since thay can land on DPC ground paved with concerete in the middle of a task force. :D
 
Yeah, Squiffy, MOIs get tossed into the mix, too, as far as all over affects for the AIs.

Blast it. Some of it is coming back a little. Not taking down notes didn't help.
And, unfortunately, with the newer FM entrys, that doesn't help much, either.

But, for the AIs, I know the accuracy of the FM has alot of control over how they fly...and act.

Just a quicky version...with an experiance with CFS2 FM.
Besides trying to get an accurate FM, also looking for right AI actions, I fooled with several things. One gripe I had (for the AI) was the un-realistic take-off, faster, and out-climbing me.
I started playing with a few numbers. I think it was wing planform, making minor changes to it, and testing the AI.

From one extreme to another. From (the AI) easily out-speeding and climbing me....to (during one test) the AI actually did take-off, but was un-able to climb!
The AI simply flew along at minimum altitude the entire flight.

Yeah, was able to adjust until the AI would take-off, with more realistic climb and speed. I got to the point, after I took off, I would have to slow a little, to let the AI catch up.
All other overall flight for the AI greatly improved.

But, I've said all this before.
 
Sure, I've wonderd about "default vertical speed" when it's set to 1000ft/m. Makes me wonder, is this their default climb and descent rate? If so, that might be why they drop so steeply. never compared it to better landing ai.
 
You must log in or register to reply here.
Back
Top