Jump to content

  • Log In with Google      Sign In   
  • Create Account

Banner advertising on our site currently available from just $5!


1. Learn about the promo. 2. Sign up for GDNet+. 3. Set up your advert!


aircraft landing


Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.

  • You cannot reply to this topic
14 replies to this topic

#1 a2k   Members   -  Reputation: 122

Like
Likes
Like

Posted 14 July 2002 - 08:25 PM

is modelling aircraft landing difficult? i''ve just been recalling collision detection algorithms causing "shaking" if trying to implement resting contacts, on the ground. i was wondering how they are handled in today''s flight sims. a2k

Sponsor:

#2 Gee   Members   -  Reputation: 122

Like
Likes
Like

Posted 14 July 2002 - 09:37 PM

did you mean:
"is simulating an aircraft landing difficult?"

or?

#3 Strife   Members   -  Reputation: 374

Like
Likes
Like

Posted 15 July 2002 - 04:46 AM

I''m sure that if you wanted accurate Bournulli''s Principle simulations, as well as gravity and everything else, it could be pretty difficult, but if you do a greatly simplified version, it can''t be too difficult



#4 a2k   Members   -  Reputation: 122

Like
Likes
Like

Posted 15 July 2002 - 06:02 AM

quote:
Original post by Gee
did you mean:
"is simulating an aircraft landing difficult?"

or?


the terms "flight model" and "physics model" are what i was referring to.

a2k

#5 Anonymous Poster_Anonymous Poster_*   Guests   -  Reputation:

Likes

Posted 15 July 2002 - 06:19 AM

If you''ve modeled the physics of the airplane correctly, you''re more than half done. After all, an airplane about to land is just an airplane flying at a really low altitude, usually dirtied up (flaps extended) to lower approach speeds.

The other thing you''ll want to simulate is ground effect. Ground effect improves the airplane''s lift efficiency, and in my experience at least is manifest at less than 10 feet or so above the runway. The result usually feels like you''re floating down the runway.

The landing happens (or rather, ideally should happen) when the aircraft no longer has sufficient airspeed to keep it aloft, allowing it to settle down onto the ground. For heavier aircraft or aircraft with violent pitch-down stall tendencies (like most low wing planes), you''d want to keep it just above stall.

Other factors you may want to throw in are gusts and crosswinds, which can definitely add some excitement to your landing (speaking from experience here!). For crosswinds, your airplane will have a tendency to yaw, fighting to turn the nose of your plane into the wind, which of course must be corrected with the appropriate rudder pedal.

Lastly, if you REALLY want to get technical, many people employ a technique known as a slip to compensate for a too-steep straight-in approach. Slip involves cross-controlling the airplane; that is, ailerons one way, rudder the other. The result is a goofy, cock-eyed attitude and a markedly accelerated rate of decent.

Hope this helps,
John Nagle
http://www.Jobie.com


#6 a2k   Members   -  Reputation: 122

Like
Likes
Like

Posted 15 July 2002 - 08:02 AM

thanks for the input AP, i''ll definitely hold your ideas in high consideration. unfortunately, the problem that i''m dealing with is related to collisions and resting contacts. for instance, if i fly a plane in IL-2, and have it skim the ground, the craft bounces around, and once the gears break, the fuselage starts skidding until it comes to a stop. now, with collision algorithms, resting contact forces are needed to prevent "bouncing around" on the ground. not sure how to implement the resting contacts.

a2k


#7 Timkin   Members   -  Reputation: 864

Like
Likes
Like

Posted 15 July 2002 - 09:17 PM

As my first flying instructor once said, "Landing is just crashing with style"!

You can either implement the collision response as an impulse based method or a constraint based method. One novel approach is to combine the two. If the resultant restituation energy is less than a chosen threshold, move from an impulse to a constraint model.

The point of my comment above is to remind you that the lift force being generated at the point of touchdown of the main undercarriage is less than the weight of the aircraft, since the wings have (at least partially) stalled. If you''ve got a decent flight model going then the lift force will be proportional to the airspeed for a given angle of attack (and subsequently coefficient of lift). So, as the aircraft rotates nose-up near the end of its decent and reduces power, it starts to wash off airspeed and thus lift. At the point of touchdown the net weight of the aircraft is still quite low, so the landing struts dont take much of a load... that''s why they don''t break unless you try and achieve touch down from 10ft up, or higher! My worst was about 5ft... and I had a sore ass after that one! As the aircraft decreases airspeed the net weight increases toward the resting mass of the aircraft.

is this the sort of information you were looking for? If not, throw this post in the trash bin and please give us some more details on what you need help with.

Cheers,

Timkin

#8 a2k   Members   -  Reputation: 122

Like
Likes
Like

Posted 16 July 2002 - 05:21 AM

yes, that's the closest i've been looking for so far. so far, the points are being responded to as collisions. in addition, a contact force is applied just to overcome the force of gravity. now, i'm not sure how to implement the "constraint" model that you've suggested. would that be the same thing as walls in quake simply preventing characters from penetrating through?

a2k

[edited by - a2k on July 16, 2002 12:24:00 PM]

#9 Timkin   Members   -  Reputation: 864

Like
Likes
Like

Posted 16 July 2002 - 12:47 PM

Yes, that''s the gist of it. There are many papers available on the web regarding impulse and constraint based collision methods. You should start with the Articles & Resources Link at the top of this page. Also check out Chris Hecker''s Annotated Bibliography.

As to more on flight mechanics, if you''ve got access to a decent library and don''t mind a technical book, I highly recommend Barnes W. MacCormick''s "Aeronautics, Aerodynamics & Flight Mechanics".

Cheers,

Timkin

#10 a2k   Members   -  Reputation: 122

Like
Likes
Like

Posted 16 July 2002 - 06:37 PM

actually, my physics model is already based on hecker''s demo, not to mention baraff''s papers as well. it''s just the collisions thing that are really getting to me, because papers i''ve read on collisions always discuss single point collisions, but not multiple collisions. baraff has been the closest to deal with this topic with resting contact, but his solution discusses a complex topic with only a few paragraphs.

i''ll dig deeper.

a2k

#11 grhodes_at_work   Moderators   -  Reputation: 1361

Like
Likes
Like

Posted 17 July 2002 - 06:47 AM

quote:
Original post by Timkin
As my first flying instructor once said, "Landing is just crashing with style"!



Do any flying these day? What type of aircraft?

Graham Rhodes
Senior Scientist
Applied Research Associates, Inc.

#12 Timkin   Members   -  Reputation: 864

Like
Likes
Like

Posted 17 July 2002 - 04:21 PM

quote:
Original post by grhodes_at_work
Do any flying these day? What type of aircraft?



Not for a long time unfortunately.... (that stupid PhD thingymajig)... and I had to let my license lapse when I went back to being a poor student all those years ago. I keep my withdrawal symptoms in check by building and flying R/C gliders and powered aircraft, but it isn''t quite enough. I''m trying to convince my wife to come learn gliding with me, since I''m of the belief that it is the ultimate in flying.

I trained on a Piper Tomahawk and flew that, and a couple of Cessnas (150 and 172 I think they were... but it''s been a long time) for several years. Unfortunately another student put the Tomahawk on its back in a nearby paddock and that substantially cut the flying schools fleet (by 33%)! After that I flew less until eventually I just stopped going (fuel prices got ridiculous).

Today I''m just one of those poor souls who gets that itch every time he gets on a commercial aircraft!



Timkin

#13 Eric   Members   -  Reputation: 138

Like
Likes
Like

Posted 17 July 2002 - 07:12 PM

Hey a2k, how''s Hecker''s physics code working for you? I''ve been working on a rigid body simulator from his tutorial over the past few weeks, for use in a racing game. Just today I''ve gotten to the point where I can steer my luge/sled-type vehicle through my half-pipe track.

Anyway, remember "Hover", an old project of mine? Didn''t we discuss its physics a long time ago here on GameDev? My vehicles floated on springs, and I thought I''d suggest this to you as a hack to avoid actually doing any multi-point collision resolution. Ever since I worked on Hover, I''ve always thought that a vertical spring would be an easy way to approximate the behavior of a wheel under suspension. You''d still *render* an appropriately-positioned wheel and suspension arm, but these components wouldn''t actually exist as bodies in your simulation.

#14 a2k   Members   -  Reputation: 122

Like
Likes
Like

Posted 17 July 2002 - 07:35 PM

okay, i''ll try it out. i''m just very caught up in getting the physics really right regarding collisions and contacts, that i''m probably just gonna kill myself doing it. it''s time to hack at my physics system to get it to play well and feel right, so hopefully the springs method will be best.

i remember hover being "the" big hit on gamedev. i never got it to fully work, but it was a good game. good to hear that the veterans are still on this board.

a2k

#15 Timkin   Members   -  Reputation: 864

Like
Likes
Like

Posted 17 July 2002 - 07:46 PM

quote:
Original post by a2k
because papers i''ve read on collisions always discuss single point collisions, but not multiple collisions.


Okay, I think we''re finally getting to the heart of understanding the specific problem you are trying to solve.

Let me see if I have this straight.

You know how to implement single point collision detection and collision response, but you''re not sure how to do multiple point detection and response... correct?

Assuming this is correct, there are two ways that I can come up with off the top of my head to deal with this situation.

Consider the bank angle of the aircraft at the point of contact with the ground. If this angle is not zero then one wheel will touch before the other.

Method a) Treat all angles less than epsilon (where epsilon is a small number) to be the same as zero and then treat your main undercarriage as a single contact point.

method b) For angles larger than epsilon, you can deal with the two contacts individually. The first contact will cause the aircraft to rotate onto the other gear since gravity will be countered (at least partially) by an off centre contact force. You then have a rotational dynamics problem to solve, but it shouldn''t be too difficult.

If the pilot is particularly bad and happens to drop the nose wheel onto the ground before getting the aircraft settled on the main gear, then just deal with the same rotational problem as method b). If they happed to ONLY get the nose wheel onto the ground, it''s going to break since they''re most likely flying INTO the ground!

I hope this helps further. If my assumption above is incorrect, could you please elaborate on the particular difficulty you are having.

Cheers,

Timkin






Old topic!
Guest, the last post of this topic is over 60 days old and at this point you may not reply in this topic. If you wish to continue this conversation start a new topic.



PARTNERS